Researching livelihoods and services affected by conflict ...€¦ · survey. Fourth, except in Sri Lanka, respondents have overwhelmingly negative perceptions of local and, particularly,

Working Paper 31Richard Mallett, Jessica Hagen-Zanker, Rachel Slater and Georgina SturgeApril 2015

Surveying livelihoods, service delivery and governance: baseline evidence from the Democratic Republic of Congo, Nepal, Pakistan, Sri Lanka and Uganda

Researching livelihoods and services affected by conflict

ii

About us

Secure Livelihoods Research Consortium (SLRC) aims to generate a stronger evidence base on how

people in conflict-affected situations (CAS) make a living, access basic services like health care,

education and water, and perceive and engage with governance at local and national levels. Providing

better access to basic services, social protection and support to livelihoods matters for the human

welfare of people affected by conflict, the achievement of development targets such as the Millennium

Development Goals (MDGs) and international efforts at peace building and state building.

At the centre of SLRC’s research are three core themes, developed over the course of an intensive one-

year inception phase:

State legitimacy: experiences, perceptions and expectations of the state and local

governance in conflict-affected situations

State capacity: building effective states that deliver services and social protection in

conflict-affected situations

Livelihood trajectories and economic activity in conflict-affected situations

The Overseas Development Institute (ODI) is the lead organisation. SLRC partners include the

Afghanistan Research and Evaluation Unit (AREU), the Centre for Poverty Analysis (CEPA) in Sri Lanka,

the Feinstein International Center (FIC, Tufts University), Focus1000 in Sierra Leone, the Food and

Agriculture Organization (FAO), Humanitarian Aid and Reconstruction of Wageningen University (WUR) in

the Netherlands, the Nepal Centre for Contemporary Research (NCCR), and the Sustainable

Development Policy Institute (SDPI) in Pakistan.

Secure Livelihoods Research Consortium

Overseas Development Institute

203 Blackfriars Road

London SE1 8NJ, UK

T +44 (0)20 7922 8221

F +44 (0)20 7922 0399

E [email protected]

W www.secureliveilhoods.org

Disclaimer: The views presented in this paper are those of the

author(s) and do not necessarily represent the views of DFID,

the EC, Irish Aid, SLRC or our partners, SLRC Working Papers

present information, analysis on issues relating to livelihoods,

basic services and social protection in conflict-affected

situations. This and other SLRC reports are available from

www.securelivelihoods.org. Funded by DFID, the EC and Irish

Aid.

Readers are encouraged to quote or reproduce material from

SLRC Working Papers for their own publications. As copyright

holder, SLRC requests due acknowledgement and a copy of

the publication

iii

Contents

About us ii

Tables and figures iii

Executive summary v

1 Introduction 1

2 Background, objectives and analytical frameworks 3

2.1 Situating the survey within the research programme 3

2.2 Objectives of the panel survey 4

2.3 Analytical frameworks 5

3 Research methodology 16

3.1 Research methodology 16

3.2 Sampling methods and description of sample 17

3.3 Survey comparisons 19

3.4 Econometric analysis 20

4 Which factors influence people’s livelihoods and wellbeing? 23

4.1 A description of livelihood activities of the sampled populations 23

4.2 Examining the wellbeing hypotheses 26

4.3 What do these findings tell us about livelihoods and wellbeing in

conflict-affected places? 29

5 Which factors influence people’s access to and experience of services,

social protection and livelihood assistance? 31

5.1 Overview of access to and experiences with services, social protection and

livelihood assistance 31

5.2 Examining the access to and experience of services/transfers hypotheses 35

5.3 What do these findings tell us about access to and experience of basic services,

social protection and livelihood assistance in conflict-affected places? 38

6 Which factors explain people’s perceptions of the state? 40

6.1 Overview of perceptions of government 40

6.2 Examining the government hypotheses 42

6.3 What do these findings tell us about people’s perceptions of the state? 46

7 Conclusions and initial policy implications 49

References 54

Annex 1: Comparative regression tables 61

Annex 2: Descriptive tables 108

iv

Tables and figures

Tables

Table 1: Composition of coping strategies index, from survey instrument 5

Table 2: Satisfaction with components and overall quality of schooling, taken from Sri Lanka survey

instrument 9

Table 3: Information on fieldwork 17

Table 4: Criteria used to select fieldwork locations 17

Table 5: Description of the samples 18

Figures

Figure 1: SLRC survey comparisons across countries: A hypothetical example of monthly income

data 20

Figure 2: Households receiving main income from subsistence farming, by country 23

Figure 3: Mean number of income sources per household, by country 24

Figure 4: Percentage of households with only one income source, out of all households with any

income sources, by country. 24

Figure 5: Main source of household income, DRC (% of households) 25

Figure 6: Main source of household income, Sri Lanka (% of households) 25

Figure 7: Main source of household income, Nepal (% of households) 25

Figure 8: Main source of household income, Uganda (% of households) 25

Figure 9: Main source of household income, Pakistan (% of households) 25

Figure 10: Distance to health clinic used (measured by time taken for a return journey), by country 32

Figure 11: Distance to primary school used (measured by time taken for a return journey), by country 33

Figure 12: Distance to water source used, (measured by time taken for a return journey), by country 33

Figure 13: Households receiving social protection and livelihood assistance, by country 33

Figure 14: Overall satisfaction with health facility used (% of respondents) 34

Figure 15: Overall satisfaction with primary school used 34

Figure 16: Perception of water quality, by country 34

Figure 17: Transfer had any positive impact, by type of transfer and across countries 35

Figure 18: Comparisons of perceptions of local and central government 41

v

Executive summary

In 2012/13, the Secure Livelihoods Research Consortium designed and implemented the first round of

a panel survey in five fragile or conflict-affected countries, generating cross-country data on people’s

livelihoods, their access to and experience of basic services and their perceptions of government. This

paper synthesises the findings of the DRC, Nepal, Pakistan, Sri Lanka and Uganda baseline surveys,

which were delivered to a total of 9,769 households in September-October 2012 (for DRC, Nepal,

Pakistan and Sri Lanka) and in January 2013 for Uganda.

Drawing on sets of descriptive statistics and regression analyses run at country level, the analysis here

sets out to identify key trends, similarities and differences across the five countries, in relation to three

broad lines of enquiry:1 factors associated with better/worse levels of wellbeing and ‘livelihood

outcomes’ (food security, asset ownership); factors associated with having better/worse access to basic

services or transfers (health, education, water, social protection, livelihood assistance); factors

associated with better/worse satisfaction with these services; and factors associated with holding more

positive/negative attitudes towards government.

Six key findings emerge when we look across the five analyses, which – to some extent – challenge

some of the received wisdom that shapes development efforts in fragile and conflict-affected areas. In

reading them, we should keep in mind that they provide just a snapshot at one moment in time. In

2015, we will be going back to the same respondents and interviewing them for the second round of

the panel survey. At that point, we will be able to measure changes in people’s livelihoods, their access

to services and their perceptions of government over time, and will therefore be able to make stronger

statements about the factors that possibly determine these.

First, the findings point to the importance of education: households with more highly educated heads

consistently have better wellbeing outcomes in terms of both household wealth and food security.

Findings suggest primary education makes a difference, but recipients of secondary education clearly

have even higher wellbeing outcomes. After the second wave of the panel survey we will be able to

unpack the causal mechanisms behind these links.

Second, we find that coverage of livelihood assistance consistently reaches less than one-third of

households – ranging from 15% to 32% of sampled households across the countries. This initial finding

suggests that, contrary to what might be expected, the transition from conflict to post-conflict does not

appear to result in renewed efforts to support livelihood rehabilitation. There is an apparent gap in both

effective strategies and effective programmes at sufficient levels of scale to support processes of

livelihood recovery, provide social protection and stimulate employment and growth. The second round

of data collection will expand our understanding of what factors affect households’ participation in such

programmes.

The third finding, and perhaps one of the most unexpected, is that levels of satisfaction with basic

services are generally quite high. This does not necessarily mean that households are accessing high-

quality services: we are measuring self-reported satisfaction levels based on respondents’ experiences

of the service. Three possible explanations are: 1) respondents may be expressing positive perceptions

owing to a social desirability bias (the tendency of respondents to give the responses others may expect

of them); 2) there may be actual post-conflict improvements in services (services may be patchy and of

low quality but still better than before); and 3) information asymmetries (people not knowing what is

available) might mean people are satisfied with limited services. We will be able to fully explore the

1 Hyperlinks take you to the relevant chapters.

vi

relevance of changing levels of access on satisfaction with services after the second round of the

survey.

Fourth, except in Sri Lanka, respondents have overwhelmingly negative perceptions of local and,

particularly, central government. For instance, only between 4% (Pakistan) and 36% (Uganda) felt the

central government cared about their opinion. Except in Sri Lanka and Uganda, at least two-thirds felt

the priorities of local government never or almost never reflected their own. These findings do not

necessarily suggest central government is doing a ‘worse job’: local government organisations are in

theory more closely connected to local populations, even if they fail to provide much in the way of

tangible development gains, and this proximity/visibility may explain the more positive perceptions. It

may be the case that the greater physical and hierarchical distance between people and the central

layer of government potentially has a limiting effect on its capacity to build legitimacy. Or, in other

words, households may associate public goods delivery with whichever level of government is most

visible to them, regardless of who delivered them, and their perceptions are shaped accordingly.

The fifth – and strongest – finding is that an individual’s overall satisfaction with a service or transfer

depends heavily on how well specific aspects of that service are run; access to the service (in terms of

distance to a service or transfer receipt) is largely irrelevant. For example, respondents’ experience with

basic services suggests factors such as ‘satisfaction with the availability of medicine’ and ‘satisfaction

with the number of teachers’ are strongly and positively associated with higher levels of overall

satisfaction with those services. Quality, as opposed to simple presence, is the most important factor.

Furthermore, whether or not someone has access to a particular service, social protection transfer or

type of livelihood assistance does not appear to matter much in explaining perceptions of the

government. Instead, we see presence of grievance mechanisms and possibilities for civil participation

strongly influences perceptions of the government – even if these are not effective in practice. Taken

together, these findings suggest there is potentially something about the way in which public services

can act as a channel through which citizens and public authorities interact (Van de Walle and Scott,

2011).

The final finding is that our data show it is not all about conflict. The five countries are characterised by

a history of conflict, yet historical/current displacement and recent sporadic conflict do not affect

wellbeing outcomes, access to services or perceptions of the state in any consistent way. They are also

characterised by multiple causes of vulnerability: conflict-related shocks can affect livelihoods,

wellbeing outcomes and perceptions of the government, but we find ‘non-conflict-related’ shocks, such

as environmental stresses, natural disasters and health or economic shocks often have equally as

dramatic effects on these outcomes. We do find evidence of long-term negative effects of serious

(conflict-related) crimes in Uganda, yet in all countries our findings are forcing us to re-examine ‘conflict

exceptionalism’. Overall, the baseline findings point to the importance of recognising and reflecting in

policies and programming, that: 1) the lives of people in fragile and conflict-affected areas are not solely

defined by their experiences of war – their livelihood choices and wellbeing are not determined by

violent conflict in an absolute sense; and 2) it is necessary to account for the longevity of the effects of

conflict, violence and insecurity in forward-looking development and recovery efforts. Findings from our

surveys suggest no simple relationship between access to services and people’s perceptions of the

legitimacy and performance of government. For donors, this suggests a need to be more cautious about

claims that supporting services contributes to state-building. That said, it is still absolutely appropriate

to support service delivery from a rights- and needs-based perspective: services are critical to

eliminating poverty, irrespective of whether they contribute to state legitimacy. But if donors want to

support state capacities to deliver services, they may need to pay much more attention to how they are

supporting services and be more explicit about how state capacities are being built at what level.

1

1 Introduction

As a multi-year, cross-country research programme, one of the overarching aims of the Secure

Livelihoods Research Consortium (SLRC) is to contribute towards a better understanding of what

processes of livelihood recovery and state-building look like following periods of conflict and how

positive outcomes are achieved. Understanding socio-economic change of this nature is only possible

when appropriate evidence exists. This, in turn, requires the availability of reliable longitudinal data that

are able to measure shifts, fluctuations and consistencies in the performance of a given unit of analysis

(an individual, a household, an economy, and so on) against a set of outcome indicators between at

least two points in time.

In order to directly address this need for appropriate evidence – evidence which tells us something

about processes playing out over time and in more than a single context – SLRC are carrying out panel

surveys in five countries: the Democratic Republic of Congo (DRC), Nepal, Pakistan, Sri Lanka and

Uganda. Designed to produce information on a range of issues – including: people’s livelihoods

(income-generating activities, asset portfolios, food security, constraining and enabling factors within

the broader institutional and geographical context); their access to basic services (education, health,

water), social protection and livelihood assistance; exposure to shocks and coping strategies; and their

relationships with governance processes and practices (participation in public meetings, experience

with grievance mechanisms, perceptions of major political actors) – the surveys are being implemented

twice in each country. The first round took place in late 2012 to early 2013, and the second round,

where we will attempt to re-interview the same respondents, will take place in 2015.

This paper summarises and synthesises the findings of the DRC, Nepal, Pakistan, Sri Lanka and Uganda

surveys. The surveys took place in September-October 2012 (for DRC, Nepal, Pakistan and Sri Lanka)

and in January 2013 for Uganda. The smallest country sample size was 1,259 respondents (DRC), and

the largest 3,175 (Nepal). It constitutes, in effect, the synthesis baseline report, to be followed up by a

subsequent report in 2016 when the second round of the panel survey is complete. This report

complements the five other country papers, published separately.

The SLRC conducted a similar study in South Sudan between March and April 2012 which preceded the

baseline for the other five countries. It was conducted jointly with the UN’s Food and Agriculture

Organisation (FAO), which is implementing the “Sustainable Food Security Through Community-Based

Livelihood Development and Water Harvesting” project (SFLDP). Their baseline survey is in essence a

livelihoods survey, to which we added questions about perceptions of government. As the baseline for

the SFLDP project, the survey for South Sudan is somewhat different in content and in objective from

the other SLRC surveys, and we have therefore not included the South Sudan findings in this synthesis

report. A summary of the findings from the South Sudan survey can be found at the end of the report.

The findings presented here are intended as tentative, preliminary ideas about the forces affecting

people’s livelihoods and wellbeing, access to and experiences of services, and perceptions of

government in these conflict-affected countries. The second wave of the panel survey, which asks the

same respondents the same questions after a three-year interval, will put these findings into context

and enable us to understand better the processes which underpin the findings. The research will also

be complemented with in-depth qualitative research in the five countries.

The paper is structured as follows. Section 2 provides background to the survey, situating the panel

survey in relation to the overarching themes of SLRC’s research programme, outlining the objectives of

the survey, and presenting the analytical frameworks and hypotheses used to guide the analysis of the

survey data. Section 3 presents the survey methodology of the in greater detail, discussing the specific

sampling methods used and describing how we have made comparisons across countries. Sections 4-6

constitute the analytical core of the paper, respectively exploring: which factors influence livelihood

2

status; which factors influence people’s access to and experience of services and social protection; and

which factors influence people’s perceptions of government. Finally, Section 7 concludes and lists some

preliminary policy implications.

3

2 Background, objectives and analytical

frameworks

This section is split into three parts. The first provides some background to the survey by situating it in

relation to the SLRC’s broader research agenda. The second outlines the objectives of carrying out a

panel survey. The third describes the basic analytical frameworks used to analyse the survey data.

2.1 Situating the survey within the research programme

The cross-country panel survey is of direct relevance to the first and third themes of SLRC’s six-year

global research programme:

1 Legitimacy. What are people’s perceptions, expectations and experiences of the state and of

local-level government? How does the way services are delivered and livelihoods are

supported affect people’s views on the legitimacy of the state?

2 Capacity. How do international actors interact with the state and local-level government

institutions? How successful are international attempts to build state capacity to deliver social

protection, basic services and livelihood assistance?

3 Livelihood trajectories. What do livelihood trajectories in conflict-affected situations tell us

about the role of governments, aid agencies, markets and the private sector in enabling

people to make a secure living?

Legitimacy: people’s perceptions of government and the role of service delivery 2.1.1

Establishing, building or strengthening state legitimacy is a major element of state-building. The OECD

(2010: 3), for example, notes that, ‘State legitimacy matters because it provides the basis for rule by

consent rather than by coercion.’ For donors, while the steps they can take to influence state legitimacy

are few, they do have an interest in developing a clearer understanding of the following: what leads to

legitimacy? What, if anything, can they do to strengthen state-society relations? And what might be the

(unintended) positive and negative consequences of their programming on state legitimacy if they, for

example, route development funding via bodies other than the formal organs of the state?

Because basic services represent a material expression of the social contract that is thought to exist

between functioning states and empowered citizens, there has been a ‘striking trend toward framing

the provision of vital public services – including health, education, water, and sanitation – as a key

source of legitimacy’ (Mcloughlin, 2014: 2). However, literature reviews carried out during SLRC’s

inception year found very little evidence for the frequent assertion that improving access to services

and social protection in conflict-affected situations contributes to state-building (see, in particular,

Carpenter et al., 2012). The relationship between delivering services and state-society relations

remains poorly understood. Given the cited importance of legitimacy in state-building processes – as

the European Report on Development (2009: 93) notes, ‘State-building efforts are bound to fail if, in

strengthening institutional capacities, the legitimacy of the state is not restored’ – it is both surprising

and concerning that we have so little robust knowledge about what leads to state legitimacy.

Despite these gaps, state-building, encompassing both legitimacy and capacity, provides the organising

framework for much international engagement in conflict-affected situations. In tackling this question,

we are thus taking up the OECD’s call for donors to ‘seek a much better understanding – through

perception surveys, research and local networking – of local people’s perceptions and beliefs about

what constitutes legitimate political authority and acceptable behaviour’ (OECD, 2010: 55).

Livelihood trajectories: tracking change and identifying determinants 2.1.2

Literature reviews carried out during SLRC’s inception year identified empirical and longitudinal

research on livelihoods in conflict-affected situations as a key evidence gap. Although sometimes good

4

in-depth case studies can be found on livelihood strategies in particular contexts, these are usually just

snapshots. Qualitative case study approaches are also insufficiently linked to quantitative survey data.

The reviews also revealed a significant gap in any comparative analysis of the effectiveness and impact

of interventions to support livelihoods (see, in particular, Mallett and Slater, 2012). There is some

evaluation and academic literature that examines the impact of particular projects or programmes, but

very little which looks at the overall significance of aid in people’s livelihoods and compares the impact

of different approaches. SLRC’s research programme aims to fill some of these gaps by building a

picture of how people make a living in particular contexts and track how this changes over time.

2.2 Objectives of the panel survey

The panel survey will help us answer parts of our research questions appearing under the first and third

themes of the research programme.

Regarding the first theme on legitimacy, our approach is centred on documenting and analysing

people’s views of government (or the formal state) in conflict-affected situations. In all countries, we

collected information on respondents’ perceptions of at least two levels of government: local and

central. In the case of DRC, information on a wider range of actors was collected, including informal

state or non-state agents (e.g. local kings or mwami in DRC). A cross-country panel survey incorporating

perception-based questions enables an analysis of people’s relationship with state authorities, allowing

us to investigate difficult-to-measure, subjective issues such as trust and satisfaction, and providing

both a comparative snapshot and a longitudinal perspective.

Under the third theme (livelihood trajectories), SLRC is undertaking rigorous, longitudinal livelihoods

research. Our aim is to build a picture of how people make a living in particular contexts, track how this

changes over time, and shed light on what causes change. We want to know whether people are

recovering or starting to build stronger and more secure livelihoods, are stuck in poverty or are sliding

into destitution, and how this is affected by the broader political, economic and security environment.

Implementing a panel survey that captures both the dynamics and determinants of people’s livelihoods

enables this.

The SLRC cross-country panel survey therefore combines elements of both perception and livelihoods

surveys, enabling a dual focus on government and legitimacy on the one hand and livelihood

trajectories on the other. There are five points of added value of conducting a hybrid survey of this kind:

1 It allows us to link perceptions directly with experiences

2 It generates rare panel data in fragile and conflict-affected contexts

3 It allows us to identify similarities and differences between different fragile state contexts

4 It allows us to differentiate between levels of government and different forms of government

5 It generates information on livelihoods beyond simple income measures.

Given the overarching objectives of the panel survey study outlined above, the more specific research

questions are as follows:

Which factors influence livelihoods and wellbeing?

Which factors influence access to basic services, social protection transfers and

livelihood assistance?

Which factors influence of experience of basic services, social protection transfers and

livelihood assistance?

Which factors influence perceptions of local and central government?

5

The next section outlines the analytical frameworks that form the basis of the survey and subsequent

analysis. We also give a basic set of hypotheses that relate to these frameworks.

2.3 Analytical frameworks

Three basic analytical frameworks emerged from the survey design process, which are outlined below. It

should be emphasised that, because this paper is based on the first round of the survey, the analysis is

not geared towards identifying and explaining changes over time (which is why we talk about livelihood

status as opposed to trajectory throughout the report). Understanding change will be possible after the

second round. Much of the analysis here focuses on producing descriptive baseline statistics and

identifying possible correlations and relationships between different sets of factors. The data collected

also allows us to explain variations between households across a range of outcomes.

Livelihoods and wellbeing 2.3.1

Livelihoods and wellbeing are broad concepts and cannot be meaningfully captured by a single

indicator. We measure livelihoods by asking about the activities, whether paid work or otherwise, that

sustain the household. We seek to measure the (changing) importance of specific activities and the

access to better livelihood opportunities that particular groups, households or people have. Wellbeing is

more complex. We have chosen to measure it in two different ways by looking at:

Food security (using the Coping Strategies Index)

Household asset ownership (as a proxy for wealth).

The Coping Strategies Index / food insecurity index is a tool for measuring current food access: the

higher the food insecurity index the worse-off the household (Maxwell and Caldwell, 2008). Five coping

strategies and their relative severity have been identified to be generally internationally applicable and

can be seen as proxies for food insecurity (Maxwell and Caldwell, 2008). The overall score of the

insecurity index for each household is calculated by multiplying the number of times in the past week

(for Uganda) or past month (for other countries) that each coping strategy was used by the severity of

the coping strategy, and summing the products. The final index score is a weighted sum reflecting the

frequency with which households have adopted particular behaviours over the course of the previous

30 days. These behaviours are given in Table 1, which replicates the survey question. Even though the

food insecurity index was measured in exactly the same way in all countries, we will not be comparing

average scores across countries, as the survey was conducted in different seasons. Instead we will be

comparing the factors that correlate with changes in the food insecurity index.

Table 1: Composition of coping strategies index, from survey instrument

In the past 30 days, if there have been times when you did not have

enough food or money to buy food, how often has your household had

to:

Only one response allowed:

1. Never

2. Rarely (once or twice in the past 30 days)

3. Sometimes (three to ten times in the past 30 days)

4. Often (more than ten times in the past 30 days)

5. Always (every day)

a. Rely on less preferred and less expensive foods?

b. Borrow food, or rely on help from a friend or relative?

c. Limit portion size at mealtimes?

d. Restrict consumption by adults in order for small children to eat?

e. Reduce number of meals eaten in a day?

The second outcome indicator, household wealth, is proxied by the assets owned by the household

using the Morris Score Index (Morris et al., 1999). The Morris Score Index is a weighted asset indicator

that weights each durable asset owned by the household by the share of households owning the asset.

What this essentially means is that households are considered better-off when they own assets not

owned by most households in the sample. The Morris index includes all productive, household and

livestock assets and included different assets in the different countries. The index has been shown to

6

be a good proxy of household expenditure in rural Africa (ibid) and has been used in many other settings

too, for example in transition countries like Albania (Hagen-Zanker and Azzarri, 2010).

Of course, it is also likely that relationships may exist between asset ownership and food security, our

respective proxies for livelihood status and wellbeing. For example, while Tschirley and Weber (1994)

find that, in previously war-affected parts of Mozambique, landholdings constituted a key determinant

of a household’s calorie consumption, across the border in southern Zimbabwe, Scoones (1995)

reports strong correlations between wealth rankings and livestock ownership, farm asset holdings and

crop harvests. Further afield, Takasaki et al. (2001) observe strong associations between levels of

household wealth and the kinds of livelihood activities engaged in by households in rural Peru.

Having been through a lengthy process of expert consultation and thorough deliberation, we propose

that variations in livelihoods and wellbeing can be explained, at least in part, by the sets of factors

outlined below. Some basic hypotheses related to these factors are listed at the end of this sub-section.

i. Household factors. Including demographic characteristics of the household, the dependency

ratio, the religion and ethnicity of the household, and education, displacement status and

migration characteristics.

We know from the existing literature that household composition and identity – and what the

individuals within them do – often affect livelihood outcomes at the individual and household

level (see de Waal and Whiteside, 2003 and Baulch and McCulloch, 2002 on the role of

dependency ratios, Meagher, 2005 on the role of ethnicity, and Kennedy and Peters, 1992 on

the role of the gender of the household head), as well as particular defining characteristics such

as adult educational attainment – which has, for instance, been found to influence both food

security indicators (Garrett and Ruel, 1999) as well as household income levels (Moser, 1998)

– and migration and displacement (Collinson, 2009; Ellis, 2003).

While our survey focuses on livelihoods and wellbeing outcomes at the household rather than individual

level, we do recognise the importance of individual factors (especially, for example, the intersection of

age, gender and ethnicity) and explore these as explanatory variables for particular patterns of

livelihoods, asset ownership and food insecurity.

ii. Contextual factors. Including location, indicators accounting for season, the occurrence of

conflict (proxied by fighting in the local area), perceptions of safety in the neighbourhood or

when moving from place to place.

Livelihoods are influenced not only by individual and household-level attributes, but additionally

by a broader set of structural and institutional factors (Carr, 2013; De Haan and Zoomers,

2005). Previous studies have observed that livelihood outcomes are sometimes a function of

geography (Pain and Lautze, 2002), perceptions of risk in the immediate and surrounding area

(Block and Webb, 2001; Rockmore, 2011), and actual exposure to conflict. Indeed, a growing

body of evidence tells us that experiences of war and exposure to intense physical violence can

have significant detrimental effects on livelihood outcomes, sometimes for many years after the

event. For example, several studies show how experiences of conflict are closely associated with

lower educational attainment (see Mallett and Slater, 2012: 16-17), which is in turn associated

with worse livelihood outcomes in many instances, while others highlight the positive

relationship between conflict and asset loss for many households (Annan et al., 2006; Brück,

1997). However, the story is not black and white: exposure to conflict and violence has been

found, in some circumstances, to promote more altruistic behaviour (Voors et al., 2012) and to

encourage the adoption of more risky yet more profitable agricultural practices (Badiuzzaman et

al., 2011) – two behavioural changes that may well have implications for a household’s

economic recovery following war.

7

iii. Shocks experienced by a household, including natural disasters, economic shocks, as well as

crime and conflict.

The development literature suggests that experience of (short-run) shocks and (long-run)

stresses can have sometimes profound effects on the capacities of people to make a living and

on their livelihoods trajectories. Aside from the effects of exposure to episodes of conflict and

violence, as discussed above, evidence shows that livelihood outcomes at the household level

can be shaped by natural disasters such as floods or earthquakes (Kirsch et al., 2012; 2011),

by long-term health problems such as HIV and AIDS (Wafula et al., 2013) and by crimes such as

cattle rustling (Schilling et al., 2012) or sexual assault (Christian et al., 2011). The effects of

such shocks and stresses can either be direct – for example, asset loss through flooding – or

indirect, such as when the experience of a particular crime leads to attitudinal and behavioural

change amongst affected individuals, which may in turn affect activities and outcomes. As such,

we asked detailed questions on health, economic and environmental shocks and crimes in the

questionnaire and included these in the analysis.

iv. Differential access to basic services, social protection and livelihoods services and the quality

of these services / transfers.

Livelihood outcomes are dependent, to a large degree, on different forms of human and

physical capital. The supply of accessible, appropriate and timely services and assistance by

governments, NGOs, community groups or other actors can be seen, in many cases, as

necessary for ensuring the maintenance and enhancement of human and physical capital

stocks at the individuals and household level (for example, through keeping people healthy,

increasing their knowledge and capacities through education, and boosting productivity through

appropriate livelihood assistance). There is a substantial body of evidence highlighting the

protective, preventative and promotive roles that different kinds of support services can play,

from cash transfers (Blattman et al., 2013) to agricultural extension programmes (Owens et al.,

2003; Anderson and Feder, 2004) – although it must be noted that, in many instances and for

one reason or another, such interventions also often fail to produce much in the way of positive

impacts.

Nonetheless, the receipt of targeted cash transfers has been linked, in South American

contexts, to higher likelihoods of entrepreneurship (Bianchi and Bobba, 2010; Ribas, 2013)

and, in South Asia, to improved school attendance (Khandker et al., 2003). The receipt of food

aid has a positive bearing on child nutrition which also affects child labour force participation or

household livelihood strategies more broadly (Quisumbing, 2003). Agricultural extension and

road building has been associated with increased household expenditure in Ethiopia (Dercon et

al., 2009). Again, it is unlikely that access to a particular service, such as health care or

schooling will have a direct influence on livelihood outcomes such as asset ownership. Rather, a

number of mechanisms may explain why, say, poor health care might be associated with lower

household wealth or food insecurity (for example, via negative long-term effects on human

capital, on a household’s dependency ratio, on intra-household resource allocations, and so on).

For example, for households in remote areas, even renting a bicycle to enable treatment-

seeking can require households to borrow money or sell assets and food stock, affecting

household wealth and food security (Obrist et al., 2007). The aim of the quantitative analysis is

to estimate if and to what extent the above factors determine the main outcome (wellbeing

status).

8

Access and experience of basic services, social protection and livelihood assistance 2.3.2

We are interested in which factors determine access to and experience of services, social protection

and livelihood assistance. Under basic services we focused on access to the health clinic the household

uses (in DRC, respondents were also asked about their access to the nearest hospital), to the primary

school the household uses, and to the household’s main water source.

Because the survey covered a large range of services, we made use of simple, relatively blunt proxies

for access. In the case of health, education and water, we considered return journey times (in minutes)

to health centres or hospitals, primary schools and water sources. In all cases except Uganda,

respondents were asked about the distance to the boys’ and girls’ school separately (to account for the

possibility of boys and girls using different schools); the average (mean) distance was used where

appropriate. For social protection and livelihood assistance, we considered whether households had

received any form of support in the past year. The survey instrument was tailored to reflect each

country’s most widespread types of social protection and livelihood assistance programmes (for a list

and the frequency of receipt see Annex 2, Tables 21-22). Naturally, many of the programmes target

specific groups, for example Nepal’s old age allowance or Sri Lanka’s Samurdhi cash transfer

programme, so by using receipt of a transfer as our measure of access we cannot necessarily rate

‘better’ or ‘worse’ access but simply establish which factors make households more likely to be

receiving any transfers.

For satisfaction with the health service the indicator used comes from the survey question, ‘Overall, how

satisfied are you with the quality of the service on the basis of your most recent use of the clinic?’ which

invited answers on a five-point Likert scale. The same type of measurement was used for satisfaction

with the school based on satisfaction with the overall quality of the service. For both of these services,

sub-questions collected data on satisfaction with components of the service and these appear among

our covariates at the analysis stage. An example of the sub-questions on school quality is given in Table

2 below.

Box 1: Hypotheses on livelihoods and wellbeing

1 Households with better-educated members have better livelihood and wellbeing outcomes

2 Households that are or have been displaced have worse livelihood and wellbeing outcomes

3 Households that have recently experienced conflict or who are living in (perceived) unsafe

locations have worse livelihood and wellbeing outcomes

4 Households that have recently experienced a greater number of shocks and crimes have

worse livelihood and wellbeing outcomes

5 Households with worse access to basic services have worse livelihood and wellbeing outcomes

6 Households with access to social protection and livelihood assistance have better livelihood

and wellbeing outcomes

9

Table 2: Satisfaction with components and overall quality of schooling, taken from Sri Lanka survey

instrument

On a scale of 1 to 5, please rate the following:

Very satisfied = 5; Satisfied = 4; Indifferent = 3; Dissatisfied = 2; Very Dissatisfied =1;

(Difficult to say = 0)

Boys’

school

Girls’

school

a. Number of teachers

b. Quality of teaching staff

c. Teacher attendance (as in opposite to absenteeism)

d. Class size

e. Quality of school infrastructure (including presence of toilets)

f. Quality of equipment (e.g. books, desks, chairs)

g. Accessibility (travel time, quality of transport, connectivity, road quality)

h. Medium of language

i. School admission process

j. Participation in national sports and cultural events

Overall, how satisfied are you with the quality of the service?

Very satisfied = 5; Satisfied = 4; Indifferent = 3; Dissatisfied = 2; Very Dissatisfied =1; (Difficult

to say = 0)

Note: Fewer components were included in other country surveys; a similar structure was followed for health

Satisfaction with water quality was indicated by whether or not the respondent perceived the water as

clean and safe. For social protection and livelihood assistance we do not measure satisfaction directly,

but rather perception of whether or not the transfer made a positive impact on household expenditure

(for social protection) and livelihoods or agricultural production (for livelihood assistance).2 Since these

data were recorded for each transfer that the household received, and households may perceive

impacts to be different for different transfer, in each country one type of transfer was chosen for the

regression analysis, usually the one with the highest coverage.

For all services, data were captured on the provider of the service and, where appropriate, whether fees

(informal or formal) were demanded for the service. These were included among the covariates. In this

sub-section we present some basic hypotheses related to these factors.

i. Household factors. Including demographic characteristics of the household, the dependency

ratio, the religion and ethnicity of the household, and education and migration characteristics.

From the existing literature we know that household composition and the attributes of

household members can be a predictor of a household’s access to services and social

protection. For example households with higher dependency ratios (i.e. more elderly and/or

children present in the household) are more likely to access social protection services (Adato

and Bassett, 2009, offer some examples of, chiefly, child-targeted interventions; Bertrand et al.,

2003, illustrate the mechanism behind pension uptake) and many social protection

interventions are specifically targeted at households with children or elderly. While social

protection programmes are often poverty-targeted, in practice, coverage can be regressive due

to programme design or elite capture (e.g. Hossain, 2010), as is the case with many food

subsidy programmes for example (e.g. Bastagli, 2014). Wealthy households may also have

2 In Nepal, Pakistan, Sri Lanka and Uganda a livelihood transfer had a positive impact if the respondent stated that the transfer

helps in any way, as opposed to stating that ‘the transfer is too small to make a difference to my life’. The possible ways in

which the transfer helped were: ‘The transfer helps me a bit: I can buy some extra food’, ‘The transfer helps me quite a lot: we

are rarely of food anymore and I can buy some other household items’ and ‘The transfer helps me a lot: we are never short of

food anymore and I can also pay for school fees or invest in a small business’. In the DRC the question was asked differently,

so here a positive impact is identified if the respondent answered ‘yes’ to the question, ‘Did this transfer/service improve your

agricultural production/ other livelihood activity?'.

10

better access to basic services (e.g. ESCAP, 2007). Finally, the literature points to the fact that

displaced households or ethnic minority households have worse access and experiences of

basic services (e.g. ESCAP, 2007). For example, in Nepal, households from socially excluded

groups (low caste groups or ethnic minorities) are shown to have worse access to services

(UNDP, 2009; ADB, 2010). Recent evidence from India shows that even within a health

insurance programme targeted at socially excluded groups, these particular groups still face

discrimination and are less satisfied with the services provided (Sabharwal et al., 2014).

ii. Contextual factors. Including location, the occurrence of conflict, perceptions of safety in the

neighbourhood or moving to work.

It is clear that location matters in determining access to services. For instance, households in

remote areas in Nepal have worse access to basic services and services provided are of lower

quality (World Bank, 2006; DFID, n.d.). This seems to be the case more broadly (ESCAP, 2007).

Intuitively, the occurrence of fighting in the area or the perception of the area being unsafe

should have a negative effect on access to services. While there is very little evidence on this

relationship, as shown above, the limited data available tend to point towards worse access to

and experience of services. As detailed in Slater et al. (2012) based on analysis from Gates et

al. (2012), citizens in countries affected by conflict and fragility have visibly worse access to

basic services. On the other hand, there is no analysis of the impact of conflict on access to and

experience of social protection (Slater et al., 2012).

iii. Implementation and performance of basic services, social protection and livelihood

assistance. Including regularity of the provision, who provides the service, number of staff, etc.

There are huge gaps in the literature, but in general it suggests that the efficiency and

effectiveness of services affects access to and experience of basic services, social protection

and livelihood assistance. It is obvious that the regularity and timing of provision affects

satisfaction with services. With regard to social protection, recent evidence from Nepal shows

that poor implementation of the Child Grant programme both affects access to the programme

(eligible households being unable to join) and satisfaction with the programme (Adhikari et al.,

2014). Other studies also highlight the importance of timing and regularity of social protection

transfers in determining satisfaction (DFID, 2011). Studies on livelihood services show that

beneficiaries are less satisfied with services or assistance that is not provided at the time when

it is most beneficial for livelihood activities (Levine, S 2013 (pers comm)). Services and social

protection in many conflict-affected places are often delivered by non-state actors (Slater et al.,

2012), but it is not clear how this affects access to and experience of services. Finally, the way

services are implemented (the ‘daily encounters’, the ‘user-friendliness’) is likely to determine

overall satisfaction with these services, as shown in a study based on Afrobarometer data for

Africa (Bratton, 2007).

iv. Differential access to basic services (and its effects on people’s experience of those services).

Finally, we expect that distance to basic services is likely to affect experience of services. In

Tanzania, long average distances to health care facilities were found to discourage uptake of

treatment (Mamdani and Bangser, 2004) and local facilities were generally under-equipped,

resulting in low patient satisfaction. Mrisho et al (2009) cite distance to the nearest hospital

among the factors resulting in lower satisfaction with pre-natal care, along with staff shortages

and quality and availability of equipment and medical supplies. In cases where rapid treatment

is needed, such as the provision of anti-malarial drugs, the timeliness of reaching the health

facility can affect the efficacy of treatment and the patient’s satisfaction (Obrist et al., 2007). In

Tajikistan, Shemyakina (2007) finds some evidence that distance to school reduces the

11

likelihood of attendance, which suggests that it may also affect parents’ or other household

members’ satisfaction with the service or their perception of its utility. A systematic review of

evidence linking the distance from a household’s water source to the occurrence of diarrhoea

found that further distances were associated with a greater health risk (Wang and Hunter,

2010). A less recent study in Sri Lanka noted that distance to water source did not correlate

with water consumption but raised the issue of water storage in the home as a possible

determinant of contamination and, as such, perceived quality (Mertens et al., 1990). On the

whole, the literature suggests that those households who travel greater distances (i.e. have

worse access according to our proxy) to basic services have worse experiences with them.

Box 2: Hypotheses on access to and experience of services

1 Wealthier and more educated households have better access to basic services and social

protection and livelihood assistance

2 Households that are or have been displaced have worse access to, and experience of, basic

services, social protection and livelihood assistance

3 Households that have recently experienced conflict or who are living in (perceived) unsafe

locations have worse access to, and experience of, basic services, social protection and

livelihood assistance

4 Respondents who are satisfied with how a service or welfare transfer is implemented are more

satisfied with the service or welfare transfer in general

5 Households that have worse access to basic services are less satisfied with those services

12

People’s perceptions of state actors 2.3.3

State legitimacy is a complex, multi-dimensional concept. The OECD (2010) identifies four main sources

of legitimacy:

- Input or process legitimacy, which refers to the agreed rules of procedure

- Output or performance legitimacy, which is defined in relation to the effectiveness and

quality of public goods and services

- Shared beliefs, which refer to a sense of political community

- International legitimacy, which refers to a recognition of the state’s external sovereignty

In a recent paper, Teskey et al. (2012: 11) discuss this and other approaches to the conceptualisation

of what constitutes legitimacy, and conclude that:

Legitimacy can originate from either performance, including how well the government is maintain

security, creating jobs, or delivering water and sanitation services, or from process, including how the

government of the day acquired power to how inclusive it is in the process of policy-making.

Given that one of SLRC’s overarching objectives is to explore the potential relationship between service

delivery and state-building, what we are primarily interested in is the performance aspect or source of

state legitimacy. However, a focus on the process dimension of legitimacy is also important. Recent

research suggests that it is process factors themselves that help explain what (sometimes) connects

service delivery to state legitimacy (Stel et al., 2012; Wild and Mason, 2012). These studies provide

some evidence that it is the way in which services are implemented and delivered that matters when it

comes to shaping how people feel about the provider; in other words, it is not just about what is being

delivered. In order for us to be able to explore these connections through our survey work, it is

necessary to generate information on (a) how people rate the quality of what they are getting (the

performance dimension), and (b) the way in which services are delivered and decisions about provision

made (the process dimension).

These are far from objective things to measure. As Teskey et al. (2012: 11) point out, ‘Even if

“performance” can be measured objectively, for it to translate into legitimacy, it has to be perceived as

such by the population’. What this means more generally is that legitimacy is ultimately a subjective

feature (ibid).

Against this backdrop, the existing literature suggests that people’s own perceptions constitute a valid

proxy measure of state legitimacy (Carter, 2011; Herbert, 2013; Hilker and Kangas, 2011). Thus, in all

countries respondents were asked about their perception of local and central government.3 It must be

noted that although the state cannot be reduced to government alone (Boege et al., 2008; Hagmann

and Peclard, 2010), local and central government actors still constitute major – and, in most countries,

the dominant – formal state structures.

To get at these perceptions, we used two main indicators: ‘To what extent do you feel that the decisions

of those in power at the local/central government reflect your own priorities?’ and ‘Do you agree with

the following statement: the local/central government cares about my opinions?’ Much research and

deliberation went into determining which actors constituted local and central government in each

country and the country questionnaires were tailored so that it was clear to respondents which level of

government was referred to.4 These perception-based questions tell us something about performance

3 In DRC, respondents were asked about six different levels of the broader governance structure, including informal local

leaders. 4 In some countries ‘local government’ was defined in a very specified way: in DRC it referred to the Sector/ Chiefdom; in Nepal

to the local elected bodies, Village Development Committee and District Development Committee, regulated by Local Self

Governance Act 1999; in Uganda to the government at the District and sub-country level; in Sri Lanka to Provincial Councils /

Urban Councils / Pradeshiya Sabas; and in Pakistan to Union Councils, Municipal Committee, Municipal Corporation,

13

legitimacy. In order to generate information on process legitimacy, we also asked some specific

questions about participation in decision-making processes vis-à-vis local service delivery as well as

some questions about people’s knowledge and use of grievance mechanisms within the services they

access. Of course, these do not capture the full range and extent of possible process factors, but they

do take us some way towards better understanding the potential connections between service delivery

and state legitimacy. It is also worth mentioning that while these questions get at people’s views of

certain organisations and symbols of the state, they do not assess their deeper belief in the ‘idea of the

state’ (that is, whether the state is considered the normal or natural arrangement of the political

community).

Among the covariates in our ‘perception of the government’ regressions were individual and household

factors, access and experience of services, experience of shocks and conflict, and civic participation.

We outline below several hypotheses involving these factors.

i. Individual and household characteristics and contextual factors, including gender, age,

education level and livelihood status of the respondent, and household characteristics and

contextual factors as discussed previously.

We are including a set of independent variables containing individual and household

characteristics, as we know from the literature that the respondent’s background influences

perceptions of government (e.g. Christensen and Lægreid, 2005; Espinal, Hartlyn and Kelly,

2006). Previous studies have shown that men or women living in different locations have

different perceptions of government (see the work by the Asia Foundation, Afrobarometer and

Transparency International’s ‘Corruption Perception Index’). While the literature has shown that

various factors matter in explaining perceptions of government, it is divided in terms of the

nature of the impact. For instance, while we might expect wealthier households to have more

positive perceptions of government, Espinal, Hartlyn and Kelly (2006) show a curvilinear effect

between socioeconomic status and trust in government institutions for a number of Latin-

American countries, with the poor and wealthy having the most positive perceptions of

government. Other evidence has suggested that government corruption creates unfavourable

conditions for business and investment, leading to lower perceptions of state legitimacy among

the affluent (Rose-Ackerman, 2008). A recent study finds that, regardless of wealth,

membership in a civil society organisation increases the propensity to take part in an anti-

government protest (Boulding and Nelson-Núñez, 2014). Likewise, the literature is divided on

the impact of gender with some showing women have more positive perceptions of government

(see Lægreid, 1993, although not in the developing country context; and Ananth Pur and Moore,

2009, in India) and others finding the opposite effect or simply more ambivalence among

women (Bratton, 2010; Logan and Bratton, 2006). Studies on sub-Saharan Africa have found

that women are less likely to support democracy and to participate in politics, as a likely result

of disenfranchisement in the home and wider society (Konte, 2014; Branisa et al., 2013).

Hence, we will not make any a-priori assumption on the nature of the effect, but will hypothesise

that gender and other individual-level characteristics shape perceptions.

ii. Shocks and crimes experienced by the households, including natural disasters, economic

shocks, as well as crime and conflict.

As we discussed earlier, the experience of (short-run) shocks and (long-run) stresses, can have

sometimes profound effects on the capacities of people to make a living. We expect that these

shocks and crimes experienced also directly influence perceptions of government. Victims of

Metropolitan Corporation and District Council. In all cases the central government referred to the national level government

based in the capital city / seat of government.

14

crimes such as sexual and gender-based violence during conflict are less inclined to trust the

state (Barnett, 2006). Economic shocks may also be drivers of state fragility (Vallings and

Torres, 2005). In the absence of a strong central government citizens turn to informal local

actors after experiencing shocks and crimes, having no expectation that the public sector will

assist them or act in their best interest (Menkhaus, 2007; Brinkerhoff, 2005). Confidence in

government diminishes among individuals who experience a natural disaster (Arceneaux and

Stein, 2006) particularly in fragile states lacking the infrastructure for disaster preparedness.

iii. Differential access to basic services, social protection and livelihood assistance.

We expect that access to basic services, social protection and livelihood assistance positively

affects perceptions of government. However, the quality of the services being provided or the

experience in using or receiving services and social protection also affects perceptions of

government. In particular, having a negative experience is likely to affect perceptions of

government actors. A number of studies on the National Solidarity Programme in Afghanistan

have found that access to the programme have improved perceptions of government at all

levels (see for example Beath et al., 2012), despite the programme not being government

funded. Other studies have shown that access to health services affects people’s perceptions

and levels of trust in their government (see, for example, Cockcroft et al., 2011 for evidence

from Afghanistan and Rockers et al., 2012 for cross-country evidence). On the other hand, not

having access may negatively influence perceptions. For instance Osofian’s (2011) focus group

discussions showed that the Social Safety Net programme in Sierra Leone was subject to elite

capture, leading to negative impacts on citizens’ perceptions of the government.

iv. Quality of and implementation of basic services, social protection and livelihood assistance,

including satisfaction with the provision, regularity of the provision, who provides the service,

number of staff etc.

The experience and implementation of services, including satisfaction with the service, the

waiting time, regularity and costs in accessing services and social protection are likely to

determine how state government is perceived by individuals, in particular if the transfer is

government provided. Recent research from Nepal shows that weak institutional capacity of the

public sector and poor implementation practices (e.g. irregular and partial payments) undermine

perceptions of local government (Adhikari et al., 2014). Coming to the question of who provides

the service, it might be expected that perceptions of state government only improve for

government-provided programmes. But this does not always seem to be the case: household

survey data from the National Solidarity Programme in Afghanistan shows that, even where

communities are fully aware of the international origins of the money and the facilitating roles of

NGOs, they still give government credit for mobilising the assistance for their benefit (Barakat,

2009). Other studies also find positive impacts on perceptions of government of non-

government provided services (see Babajanian et al., 2014; OECD, 2011). A study by Kruk et al.

(2010) finds that equitable health care service provision can lead to higher levels of social

cohesion and perceived goal alignment between individuals and the state in the context of post-

conflict development. Using Afrobarometer data, Bratton (2012) illustrates that access to health

and education are positively correlated with satisfaction with democracy, and that so too is

satisfaction with the provision of social services. In his model, the positive effect of satisfaction

with social services is even large enough to offset the negative effect of poverty on perception of

state legitimacy.

According to proponents of the legitimacy approach to state-building, efficient public service

provision helps to secure mass confidence in government institutions and trust in the political

15

system as a whole (Inglehart and Welzel, 2005; Mishler and Rose, 2001; Klingemann, 1999).

Participation in civil society is thought to be higher among citizens who regard state legitimacy

either as very strong or very weak (a U-shaped relationship) (Seligson, 2002) and the direction

of causality may be quite fluid. Finkel et al. (2000) also find mixed results regarding the link

between civic engagement and trust in government in the Dominican Republic. We expect to

see citizens who actively participate in community meetings and make use of feedback

channels for government-provided services to hold stronger opinions on the legitimacy of

government. On the other hand, exposure to corruption in the provision of public services, such

as having to make informal payments for services, is associated with worse perceptions of state

legitimacy (Seligson, 2002; Della Porta, 2000; Morris, 1991).

The aim of the quantitative analysis is to estimate if and how much the above factors – and in particular

those relating to services – determine the main outcome (perceptions of government). We pay

particular attention to how (dis)satisfaction with services, engagement in civil society and use of

grievance mechanisms link to perceptions of government.

Box 3: Hypotheses on perceptions of the government

1 The gender and education level of respondents shape their perceptions of the government

2 Respondents living in households that have recently experienced conflict, who live in

(perceived) unsafe locations, or who have recently experienced a shock or crime have worse

perceptions of the government

3 Respondents living in households that have better access to basic services, social protection

or livelihood assistance have more positive perceptions of the government

4 Respondents who have a more positive experience in accessing basic services, social

protection or livelihood assistance have more positive perceptions of the government

5 Respondents who have access to grievance mechanisms (in the context of service provision)

have more positive perceptions of the government

6 Respondents with higher levels of civic participation have more positive perceptions of the

government

16

3 Research methodology

This section first covers parts of the survey design process, describing the research methodology,

before clarifying the sampling methods used and explaining how comparisons will be made.

3.1 Research methodology

The design of the research methodology involved a number of stages. Of particular importance was a

two-day workshop in June 2012, which was attended by selected individuals with particular experience

and expertise in survey design and implementation, including staff from the Asia Foundation, FAO, the

Igarape Institute in Brazil, as well as a number of internal SLRC partners (see SLRC, 2013). A number of

design puzzles were discussed at the workshop, before the core ODI survey team developed the

research and sampling methodologies and survey instrument further. The generic survey instrument

was then tailored to specific country contexts (see more on this below) and tested in the field in each

country. We then analysed the pilot data and revised the survey questionnaires accordingly. For more

information about the survey development process, see SLRC (2015).

In terms of survey content, a generic questionnaire was developed that was then adjusted to meet the

specific research priorities of the country teams and to fit the country context. We did not aim to

generate a system of universalist rankings between countries. Instead, the survey was designed to

allow us to identify some general trends and identify some similarities or differences between our

countries. This means that we had a number of core modules (namely the access and experience of

services modules) and some modules that were identical in all countries (notably the food security

module) to allow comparability across the different country studies (see Section 3.3 below for a more

detailed discussion on comparability). The following modules were included in all surveys: basic

household and individual information; assets; livelihood sources and activities; food security; shocks;

security shocks and justice services; basic services; social protection; livelihood services and

government.

Panel surveys are particularly rare in fragile and conflict-affected contexts. Part of the reason for this is

they are at risk of attrition – that is, when households drop out of subsequent survey rounds – and it is

assumed that because conflict often results in displacement, attrition is especially high in conflict-

affected situations. As a result, we substantially increased the sample to account for possible attrition

(see Section 3.2). The first round of the panel study was conducted in 2012 and the second round will

be conducted in 2015-2016. Attrition may be non-random; that is, some individuals with specific

characteristics are more likely to drop out of the survey. This is something we will test once we have

collected the second round of data.

The SLRC survey incorporates elements of both a livelihoods and a perception survey, which raises a

methodological issue: while the ideal unit of analysis for the livelihoods survey is at the household level,

for the perception survey it is at the individual level. Nevertheless, after extensive discussion and

consultation, the decision was reached to combine them in one survey, partly due to logistical and

budget considerations, and partly in an active effort to link perceptions more directly to real and

measurable changes in wellbeing. We opted to sample households, but to specifically seek out a varied

range of individuals within households to avoid a strong bias of male household heads for the

perception questions. Between 34% (Pakistan) and 63% (Uganda) of the respondents were female. The

decision to combine the two surveys has a number of implications and limitations (see SLRC, 2015).

Fieldwork was conducted in 2012-2013 in the locations shown in Table 3.

17

Table 3: Information on fieldwork

Country Fieldwork dates Fieldwork locations

DRC September-November 2012 South Kivu province

Nepal September-October 2012 Bardiya, Ilam and Rolpa districts

Pakistan September-October 2012 Swat and Lower Dir districts

Sri Lanka September-October 2012 Jaffna, Mannar and Trincomalee districts

Uganda January 2013 Acholi and Lango sub-regions

3.2 Sampling methods and description of sample

The sampling strategy was designed to select households that are relevant to the main research

questions as well as being of national relevance, while also being able to produce statistically significant

conclusions at the study and village level. We met these objectives by combining purposive and random

sampling at different stages of the sampling strategy. The first stages of the sampling process involved

purposive sampling, with random sampling only utilised in the last stage of the process. We selected

sampling locations purposely (including districts and locations within districts) and then randomly

selecting households within these locations. While this means that we have a rigorous sample that is

geared towards meeting the objectives of the research, the samples are not representative for the case

study countries and we cannot draw generalisations for the case countries as a whole, nor for districts.

The samples are only representative at the village level.5

Sampling locations (sub-regions or districts, sub-districts and villages) were purposively selected, using

criteria, such as levels of service provision or levels of conflict, in order to locate the specific groups of

interest and to select geographical locations that are relevant to the broader SLRC research areas and

of policy relevance at the national level. For instance, we selected locations that experienced high/low

levels of conflict and locations with high/low provision of services and tried to include locations that

accounted for all possible combinations of selection criteria. We selected survey locations with different

characteristics so that we could explore the relevance of conflict affectedness, access to services and

variations in geography and livelihoods on our outcome variables. Depending on the administrative

structure of the country, this process involved selecting a succession of sampling locations (at

increasingly lower administrative units).6 The specific criteria used in the five countries are shown in

Table 4.

Table 4: Criteria used to select fieldwork locations

Country Selection criteria

DRC Variations in livelihoods / differential access to basic services / different ethnic groups

Located in different chiefdoms (chefferie) and preferably cover more than one territory

There was a focus on rural areas

Located at different distances from Bukavu

Security considerations

Nepal Levels of conflict-affectedness

Levels of service provision

Levels of accessibility

Covering different castes and ethnicities

Pakistan Conflict affectedness and displacement of people during conflict

Interventions for rehabilitation of displaced and returnees

Sri Lanka Concentration of resettled households

Concentration of fisher households

Accessibility/security considerations

5 With the exception of Uganda, where the sample is also representative at the sub-region level. 6 For instance, in Nepal we first selected districts, then Village Development Committees within those districts.

18

Uganda Level of war affectedness

Levels of service provision

Geographical coverage varied considerably across countries. For example in Nepal and Sri Lanka the

survey covered three districts, whereas in Uganda it covered two sub-regions (totalling 15 districts), to

ensure greater policy relevance. In DRC, on the other hand, there was no demand for a broader dataset

and due to the great geographical, socio-demographic and cultural differences in the two originally

selected districts (South Kivu and Equator) any comparison between them would have been difficult

and of limited value, hence a decision was made to focus on South Kivu only. The number of research

locations varied considerably across countries (see Table 3), depending on the country-specific

sampling decisions.7

The survey did not attempt to achieve representativeness at the country or district level, but we did aim

for representativeness at the sub-district or village level through random sampling (see Table 5 for the

level of representativeness per country). Households were randomly selected within villages so that the

results are representative and statistically significant at the village level and so that a varied sample

was captured. Households were randomly selected using a number of different tools, depending on

data availability, such as random selection from vote registers (Nepal), construction of household

listings (DRC) and a quasi-random household process that involved walking in a random direction for a

random number of minutes (Uganda).

The samples are statistically significant at the survey level and village level (in all countries) and at the

district level in Sri Lanka and sub-region level in Uganda. The sample size was calculated to achieve

statistical significance at the study and village level, to accommodate the available budget, logistical

limitations, and to account for possible attrition between 2012-2015. In a number of countries

estimated population data had to be used as recent population data were not available (see SLRC

(2013).

The minimum overall sample size required to achieve significance at the study level, given population

and average household size across districts, was calculated using a basic sample size calculator at a

95% confidence level and confidence interval of 5. The sample size at the village level was again

calculated at the using a 95% confidence level and confidence interval of 5.8 Finally, the sample was

increased by 20% to account for possible attrition between 2012 and 2015, so that the sample size in

2015 is likely to be still statistically significant.

The overall sample required to achieve the sampling objectives in selected districts in each country

ranged from 1,259 to 3,175 households. The required sample sizes were achieved in all countries.

Response rates were extremely high, as shown in Table 5 (along with some other information on the

samples).

Table 5: Description of the samples

Country Sample size (#

of households)

Number of

villages

Level of

representativeness

Response rate Share of female

respondents

DRC 1,259 9 Groupement & chefferie level 98.73% 57%

Nepal 3,175 24 Ward level 99.94% 56%

Pakistan 2,114 22 Union council level 100% 34%

Sri Lanka 1,377 12 GN division & district level 100% 62%

Uganda 1,844 90 Village & sub-region level 99.94% 63%

7 The country level sampling processes are discussed in more detail in the country baseline reports, which can be downloaded from

www.securelivelihoods.org/content/2261/SLRC-Survey. 8 With the exception of Uganda, where a somewhat different sampling approach – Population Proportional to Size – was followed, resulting in

smaller clusters per village with higher confidence intervals, but being a representative sample at the sub-regional level (see Mazurana et al.,

2014).

19

As discussed in Section 3.1, in order to ensure a varied sample for the perceptions questions, we

wanted to avoid interviewing only the heads of households. Instead we aimed to interview varying

demographic groups within household. While not strictly part of the sampling (as our sampling unit of

analysis is the household), this did have implications for the fieldwork. In all countries we aimed for a

particular share of the sample to be female. This share was based on the pilot and hence took account

of local cultural sensitivities. The share ranged from 34% in Pakistan to over 60% in two countries, and

this share was achieved or exceeded in all countries (see Table 5 for the shares and SLRC (2015) for a

discussion of the difficulties of reaching respondents of a particular sex). The share of female

participants was monitored throughout the fieldwork and if the share of female participants dropped too

low, enumerators were instructed to specifically request female participants the following day. The

analyses were not weighted by gender, but we did account for the gender of the respondent, wherever

possible.

3.3 Survey comparisons

As discussed in Section 3.1 above, the surveys contained a mix of questions common to all countries

(around 80% of the survey instrument) — albeit tailored to make sense in the specific country context —

and country-specific questions. Additional modules were added for some countries to address specific

research areas of interest. For example, a module on fisheries was added to the Sri Lanka instrument

and a detailed module on serious crimes experienced by household members to the Uganda

instrument. Further, some questions had to be cut or changed in some countries because they were

culturally inappropriate (for example, some questions on specific shocks experienced by households

were dropped from the Sri Lanka instrument).

The survey design was informed by the need to produce comparative findings across countries. While

we were very clear that we wanted some degree of comparability, non-nationally representative surveys

and completely different settings meant that direct comparisons or producing aggregates across

countries did not make much sense. Hence, we did not aim to aggregate the data (and make

comparisons between, for example, women and men across all countries), nor achieve a uniformity of

indicators and values across all countries.

Context is too much of an overriding factor to allow us to aggregate and make such direct comparisons.

We instead attempted to design a survey that would allow us to identify some general trends and

identify some similarities or differences (with appropriate caveats) between our countries. As a

hypothetical example, Figure 1, below, highlights the kind of comparisons we would like to make and

the kind to avoid.

20

Figure 1: SLRC survey comparisons across countries: A hypothetical example of monthly income data

In order to be able to draw comparable findings on the analytical frameworks and hypotheses described

in Section 2, very similar regressions were run for all countries. The regression tables are listed in Annex

1.

With regard to the coping strategies / food insecurity index, even though the index was measured in

exactly the same way in all countries, we do not compare average scores across countries, as the

survey was conducted in different seasons. Instead we will be comparing the factors that correlate with

changes in the food insecurity index. The Morris Index is also not comparable between countries

because, though it includes all productive, household and livestock assets, these are different across

countries and some assets with very low levels of ownership were excluded in some countries in order

not to skew the findings.

3.4 Econometric analysis

The analysis this study is based on uses standard regression techniques, described below. As is

customary for studies drawing on regression analyses, we outline a number of limitations associated

with the econometric analysis. These caveats are standard and common to many econometric studies.

We also explain how we have mitigated any limitations.

In order to identify factors which appear to (partially) determine outcomes of various kinds – for

example, food security or perceptions of formal state actors – and compare them across countries, it

was necessary for SLRC researchers to carry out standardised regression analyses of the survey data. If

the analysis were being carried out solely at the country level, what would ordinarily happen is that each

country team would make their own decisions – based on theory, existing knowledge and context –

about which dependent and independent variables to include in each of their regressions and which

specific regression methods to use. In an attempt to generate findings which would usefully tell us

something about patterns or discrepancies across countries, it was originally decided that each country

team would include a standardised list of independent variables in each of their regressions and use

the same regression techniques; this would then enable the global survey team to produce a synthesis

based on similar looking analyses at the country level. This list of independent variables is based on the

analytical frameworks outlined above.

The nature of the dependent variable determined the type of regression run: for continuous variables,

such as the Morris Index or distance to health centre, we use an OLS regression model; for binary

21

variables, such as receiving or not receiving a social protection transfer, we used a logit regression

model; and for categorical variables, such as perception of government, we used a multinomial logit

model. A multinomial model was chosen rather than an ordered logit because although there can be

said to be an inherent order to an outcome variable where, say, ‘Dissatisfied’ equals 1, ‘Neutral’ equals

2, and ‘Satisfied’ equals 3, these evenly-spaced values do not correspond to evenly-spaced, hierarchical

unit. In particular, having a ‘neutral’ or ‘sometimes’ category is problematic for ordered logits since the

interpretation necessitates there being higher and lower categories. Neutrality may simply reflect

apathy rather than, for example, a ‘higher’ or better view of a service than that of a respondent who

claims to be ‘dissatisfied’.

Categorical independent variables were included as dummy variables. Questions only addressed to sub-

groups within the population were not included in the main regression. For instance, while we are

interested in the relationship between experience of education services and perceptions of government,

we could not include this variable in our main regression, as this would reduce the regression to

households with school-age children. Therefore we did not include any of the education variables in the

main regression and instead ran a separate regression for the parents of school-age children sub-group.

In the tests on access to education (distance to school) and satisfaction with education service, we

decided to aggregate girls and boys schooling into the same regression. Therefore in the regression on

distance to school, for example, the independent variable is the average (mean) distance to school for

boys and girls in the household. Given that the regression controls for factors relating to girls’ and boys’

schooling this means that this regression, and the regression on satisfaction with education, are run

only for households in which boys and girls attend school. In the regressions where we looked at

satisfaction with social protection and livelihoods, our dependent variable was one type of social

protection/ livelihoods assistance transfer. These are formatted red in Tables 21 and Tables 22 in the

Annex.

Independent variables were included relating to household factors, context, shocks, access to services,

and experience of services (where relevant). Under the category of household factors the regressions

controlled for the gender of the household head, average age within the household, education level of

adult members, whether the household contained a migrant, remittance receipt, displacement history,

and dependency ratio (the ratio of non-working age to working age household members), ethnicity

(ethnicity was not used in one country) and religion (religion was not used in four countries). There was

considerable variation in precisely how these variables were constructed, depending on the country

context. For example: in the case of education level, in Nepal, Pakistan and the DRC the average

education of adult household members was used; in Sri Lanka the share of adults completing primary

education was used; and in Uganda the education level of the household head was used as a

categorical variable.

Under the category of contextual factors the control variables included urban/rural context, experience

of conflict in the last three years, access to credit (used only in a few cases), experience of shocks and

crimes, perception of safety, and location. Again there was variation between how these indicators were

constructed in each country case. In Nepal and Uganda the safety variable referred to perception of

safety within the neighbourhood; in the case of Pakistan it referred to safety when moving from place to

place; in the DRC both safety at home and in the neighbourhood were included; and in Sri Lanka an

aggregate measurement of overall safety was used. We also included locational dummies (e.g. for

district) to control for differences in locations (e.g. climate, accessibility). Where variation in terms of

conflict varies by location, these location dummies may also pick up on the experience of conflict.

In the access to and experience of services categories the regression generally controlled for the

provider of a particular service, satisfaction with particular elements of the service, and formal and

informal costs associated with the service. In the regressions on perception of local and central

government responsiveness the regressions also controlled for the responsiveness and accountability

22

of service providers and for participation in community decision-making processes. These controls were

also used in certain other regressions and again differed between countries as to exactly which

indicators were used.

Following a cross-country approach, however, creates a trade-off. For instance, including a long list of

comparable independent variables means including certain variables that for some countries may be

less relevant or even collinear.9 As such, we have tested for multicollinearity in all regressions and have

re-specified those which were affected by this problem – at the expense of losing some cross-country

comparability. This explains some of the differences in variable specification discussed above.

Other reasons why the results are not completely comparable across countries include: low numbers of

responses for some questions or variables, and low levels of variation between responses for some

questions and variables. When either situation arose, such variables were not included in the

regression analysis. Some dependent variables also had to be redefined in some of the datasets as

variation or observations were so low that the originally defined model could not be run. For instance,

for some of the countries more than 90% of respondents stated that they never agree with decisions

made by the government, making it impossible to run a multinomial logit model owing to lack of

variation. In these cases, dependent variables were redefined as a categorical variable with fewer

options or as binary variables.

Finally, we would note that there is uncertainty regarding directions of causality as this is the baseline

and we are not yet utilising longitudinal data. For example, although we might find significant

relationships between variables, we cannot be sure in many cases that it is the dependent variable that

is affecting the dependent variable rather than vice versa. Whenever we are unsure of the direction of

causality, we note this and hypothesise about possible explanations for the findings. Furthermore, in

some cases a third set of factors could explain both outcomes. Again, we will note this whenever we

suspect it to be the case.

9 A scenario when two independent or explanatory variables share a strong linear relationship, which biases the results.

Box 4: Gender and intersectionality In this box we briefly outline how gender features in this report. While we recognise the importance

of the analysis of gender and other identity characteristics (e.g. ethnicity), we do not focus

specifically on them in this report – they will be the focus of the gender synthesis (SLRC, upcoming).

There we will also delve into intersectionality issues. However, variables to account for gender and

identity characteristics have been included throughout the regressions, for instance gender of the

respondent or gender of the household head. The latter has been a challenge at times, as we did

not include gender of the head in most of the questionnaires due to measurement and definition

difficulties. Instead we used proxies for the gender of the head in the analysis – for instance

defining single-parent households with one female adult as female-headed households. On the

whole, gender and other identity characteristics were not significant in the regression analysis (for a

more detailed discussion see SLRC, forthcoming).

23

4 Which factors influence people’s

livelihoods and wellbeing?

This section summarises and synthesises the analyses performed to answer the research question:

‘which factors influence people’s livelihoods and wellbeing?’ More specifically, we return to our original

hypotheses (Box 1 in Section 2.3.1), drawing on the cross-country findings to investigate the validity of

each hypothesis in turn. For the comparative analysis, we have focused on two indexes: the food

insecurity index and an asset index (Morris Score Index) (see Section 2.3.1). Additional country-specific

analyses also looked at livelihood activities and income sources.

This section is split into three parts. The first provides some basic descriptive statistics from each

country in order to illustrate key livelihood characteristics of our sample populations within each

country. The second part draws on regression findings from across the country studies to answer

whether the hypotheses presented in Box 1 (in Section 2.3.1) hold true. The third part discusses the

findings in a more narrative-based way.

4.1 A description of livelihood activities of the sampled populations

This section summarises the key findings on livelihood activities from the country reports,10 making

comparisons across countries using descriptive statistics. The focus is on understanding what livelihood

activities households participate in. We then move on to the determinants of wellbeing outcomes in the

next section.

Figures 5-9 show which livelihood activities are the main income source of sampled households. The

importance of subsistence agriculture in people’s livelihoods in immediately apparent. With the

exception of Pakistan, subsistence agriculture is the most frequent main income source for the case

study countries. Figure 2 shows that subsistence agriculture is the most important income source for as

few as 16% of households (Pakistan) to as many as 83% of households (Uganda). These findings are

consistent with other studies, which highlight the importance of subsistence agriculture in the case

study countries (see Steimann, 2005 for Pakistan; NPC, 2011 for Nepal; McDonagh et al., 2005 for

Uganda; Weijs et al., 2012 for DRC, and Fernando and Moonesinghe, 2013 for Sri Lanka).

Figure 2: Households receiving main income from subsistence farming, by country

10 For more detailed country findings, the reader should consult the country reports, as listed in the reference list.

70

46

16 31

83

0

10

20

30

40

50

60

70

80

90

DRC Nepal Pakistan Sri Lanka Uganda

Subsistence farming

24

Subsistence agriculture is the only livelihood activity where we find consistent patterns across all

countries. Other primary income sources include:

Remittances. These play an important role in two of the Asian countries: they are the most

important income source for 10% of households in Nepal and 31% of households in

Pakistan. Previous research also highlights the growing importance of migration as a

livelihood study in these countries (see Steimann, 2005 and CBS and NPCS, 2012).11

Casual labour. This is the most common income source for 11% of households in DRC, 16%

of households in Nepal, 23% of households in Pakistan and as many as 37% of households

in Sri Lanka – making it the biggest income source in the Sri Lanka sample.

Petty trade or having your own business is found mostly in Sri Lanka (14% of households)

and DRC (7%).

It has been argued that diversification (of income sources) can help households reduce vulnerability to

shocks (Ellis, 1998; 2008) but how diversified are the livelihood activities of households in our survey

samples? In general, diversification is low, with households having just over two income sources on

average (Figure 3). There is limited variation by country with somewhat higher diversification levels in

Uganda. Nearly half (47%) of sampled households in Sri Lanka and more than a third (37%) of

households in Pakistan had only one income source. However, the remainder of households in our

survey samples have more than one income source (Figure 4).

Figure 3: Mean number of income sources per household, by country

Figure 4: Percentage of households with only one income source, out of all households with any

income sources, by country.

11 Further SLRC research conducted in 2013/2014 set out to describe and explain, using mixed methods research, the multi-

dimensional process of international labour migration from two post-conflict contexts – Rolpa, Nepal and Khyber Pakhtunkhwa

(KP), Pakistan. It has shown that despite the huge financial (and sometimes physical) costs involved, international labour

migration is seen as a viable and obvious livelihoods option for those in our case study areas, largely due to the perceived

scarcity of other opportunities (see Hagen-Zanker et al., 2014).

2.3 2.1 1.8 1.7

2.9

0

0.5

1

1.5

2

2.5

3

3.5

DRC Nepal Pakistan Sri Lanka Uganda

Mean number

of

income sources

16 28

37 47

11

84 72

63 53

89

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

More than one source One source

% of households

25

The next section focuses on regression analysis through which we examine the determinants of

wellbeing.

Figure 5: Main source of household income,

DRC (% of households)

Figure 6: Main source of household income, Sri

Lanka (% of households)

Figure 7: Main source of household income,

Nepal (% of households)

Figure 8: Main source of household income,

Uganda (% of households)

Figure 9: Main source of household income,

Pakistan (% of households)

70

11

7

2 6 1

2 1

Subsistence farming

Casual labour (any)

Petty trade

Own business

Private sector (non-a

Remittances

Public sector

Other

31

37

0

14

4

4

7 4

Subsistence farming

Casual labour (any)

Petty trade

Own business

Private sector (non-a

Remittances

Public sector

Other

46

16

3

12

5

10

8

1

Subsistence farming

Casual labour (any)

Petty trade

Own business

Private sector (non-a

Remittances

Public sector

Other

83

1 2

3 4 0 0

7 Subsistence farming

Casual labour (any)

Petty trade

Own business

Private sector (non-a

Remittances

Public sector

Other

16

23

1

11 8

31

7 3

Subsistence farming

Casual labour (any)

Petty trade

Own business

Private sector (non-a

Remittances

Public sector

Other

26

4.2 Examining the wellbeing hypotheses

We are interested here in what kinds of variables are associated with better or worse wellbeing

outcomes. There are a number of things we might expect to be influential, from various household

characteristics – education levels, for example – to certain contextual features, such as (perceived)

levels of safety in the local environment. In Section 2.3.1, we outlined a set of six wellbeing hypotheses

(Box 1). These hypotheses contained statements regarding the kinds of relationships we would expect

to see between wellbeing outcomes and a range of independent variables. In this section, we draw on

findings from country-level regression analysis to examine the validity of our hypotheses. We do so by

looking across a set of regression tables in an attempt to identify patterns and consistencies (see Annex

1). In the third and final part of this section, we discuss what these findings tell us about livelihoods and

wellbeing in conflict-affected places more generally.

Hypothesis 1: Households with more highly educated members have better wellbeing outcomes

In all countries we find that an increase in the average education level of adult household members is

strongly associated with greater asset ownership (as indicated by higher MSI values), and this finding is

statistically significant.12 This positive relationship is also found when it comes to food security

outcomes: in four countries (DRC being the exception), greater levels of adult educational attainment

are associated with lower levels of household food insecurity). In the case of Uganda we control for the

education level of the household head (rather than the average in the household) and find that

households with a tertiary-educated household head are expected to have by far the highest MSI value

and the lowest food insecurity, holding other factors constant. In the case of DRC, it is worth mentioning

that, whilst not statistically significant, we still find a positive relationship between education levels and

food security. Thus, across both wellbeing outcomes – asset ownership and food security – there is

strong and consistent evidence pointing to the important role of educational attainment within our

different countries’ sample populations. Of course, we must also acknowledge the possibility of reverse

causality: this might instead be evidence that better off households are likely to be better educated –

for example because they are more able to send their members to school for longer.

Hypothesis 2: Households that are or have been displaced at some point have worse wellbeing

outcomes

Regression findings suggest that being or having been displaced at some point are not statistically

significant in explaining variations in either asset ownership or food insecurity. Only in the case of DRC

did we find any sort of clear relationship: households in the South Kivu sample population who had

been displaced were both more likely to be poorer (lower levels of asset ownership) and more food

insecure, and displacement seems to have a fairly strong link with both outcomes.

Generally speaking, however, displacement status does not appear to be particularly influential in

determining these particular wellbeing outcomes. This is surprising and we think it can be explained in

two ways. The first is that we are not, in some countries (Sri Lanka and Pakistan are the exceptions),

able to accurately record the timing of displacement, so it is possible that many other variables have

influenced people’s livelihoods and wellbeing since displacement took place that create noise in our

analysis. Second, there is some variation in how displacement was defined in the different countries. In

the DRC and Pakistan we use a variable indicating whether the household has been displaced due to

conflict (in the Pakistan case this refers to the specific conflict between 2007 and 2009). In Sri Lanka

and Uganda we use a variable indicating whether the household has ever been displaced (regardless of

12 The decision on which education indicator to include in the regressions was made independently by each partner at the

analysis stage, as a result of which the ‘average education of household members’ variable measures something different in

each country sub-sample. In DRC and Pakistan the variable used is mean years of education among adult household

members; in Nepal it is the median years of education of adult household members; in Sri Lanka it is the share of adults in the

household who have completed primary education; and in Uganda it is the education level of the respondent, in ordinal

categories, where the reference category is ‘no education’.

27

reason for displacement). In Nepal we use a variable indicating if the household has ever lived in a

different village and followed up with questions about why people moved. (We found the majority of

movements were at marriage, rather than for conflict). In Pakistan and Sri Lanka, more than 90% of the

sample had been displaced, and in Uganda and Nepal this figure was approaching two-thirds.

Hypothesis 3: Households that have recently experienced conflict or are living in (perceived) unsafe

locations have worse wellbeing outcomes

We find little evidence of a relationship between experiences of conflict (in the last three years) and

either asset ownership or food insecurity. Statistical associations are non-significant in all cases apart

from DRC, where we observe a series of negative associations: a household which has experienced

conflict recently is both more likely to own fewer assets and more likely to be food insecure. This result,

however, does not mean that we should bring to an end the discussion of how exposure to conflict

affects wellbeing and livelihoods. Our survey instrument only asked respondents to recall experience of

conflict over a three-year period which did not, in some cases, capture the period in which the country’s

last major conflict occurred (for example, in Nepal). The three-year recall period, although limiting to our

analysis at this stage, was chosen with the second panel wave in mind when three years will have

passed since the first round. In the Uganda survey, where we did collect information on long-term

exposure to serious crimes (that household members experienced during the conflict), we do find that

experiences of serious crimes are associated with lower wealth and worse food insecurity (Mazurana et

al., 2014).13

On the relationship between wellbeing outcomes and local safety (as proxied by perception-based

questions), the picture appears highly mixed. In one case for the Morris Score Index outcome and in two

cases for the food security outcome, we find that our safety variables (perceived safety in community;

perceived safety when moving around outside) are not significant at all. In others, however, we observe

inconsistent associations. For example, respondents in our Pakistan sample population who feel safe

when moving around are much more likely to live in poorer households (as measured by asset

ownership), while Sri Lankan respondents who feel safe are more likely to live in better-off households.

In Uganda, respondents who feel safe in their community are more likely to live in both poorer and more

food-insecure households, while respondents in DRC who similarly feel safe in their community are

more likely to live in better-off but more food-insecure households. In short, the regression results are

highly inconsistent, suggesting, at the very least, a non-linear relationship between (perceived) safety

and wellbeing outcomes, and one that is very much dependent on the nature of conflict and specific

contextual features.

Hypothesis 4: Households that have recently experienced a greater number of shocks and crimes have

worse wellbeing outcomes

According to the regression results, there are few consistent patterns regarding the relationship

between the number of shocks and crimes experienced and wellbeing outcomes. The survey asked

about the household’s experience during the last three years of a long list of shocks and crimes,

including health shocks, environmental disasters, economic and livelihood shocks, and crimes (for a full

list see Annex 2, Table 23). The three-year recall period was chosen for the reasons outlined in the

previous sub-section (Hypothesis 3). On asset ownership, we find that in two countries a household

experiencing a greater number of shocks in the last three years is more likely to be better off (Pakistan,

Uganda). However, this relationship is reversed in Sri Lanka, where households experiencing a greater

number of shocks are more likely to be worse off. In two countries we find a positive and statistically

significant association between the number of crimes experienced in the last three years and levels of

asset ownership (Pakistan, Sri Lanka). The influence of shocks/crimes on asset ownership therefore

13 Serious crimes include those committed by parties to the conflict that are considered as such under international

humanitarian and human rights law.

28

appears quite unclear; if anything, our evidence suggests that exposure to a greater number of

shocks/crimes may in some cases by linked to greater levels of asset ownership. This counter-intuitive

result could be a consequence of wealthier households being more likely to be targeted by criminals –

so the causality could be the other way. The second wave of the survey may allow us to uncover the

processes behind some of these ambiguous results.

The story is slightly clearer when we look at food insecurity. In three of our five countries, we find that

the greater the number of shocks experienced, the more food insecure a household is likely to be (DRC,

Pakistan, Sri Lanka). With the exception of Pakistan, we observe a similar pattern vis-a-vis the

experience of crimes over the last three years (in Uganda, this is only true for households experiencing

serious crimes; the experience of other crimes appears to be associated with the opposite effect). We

find over all in this baseline synthesis that the experience of shocks and crimes seems to be

associated with greater food insecurity (and – to a lesser extent – with greater asset ownership).

Hypothesis 5: Households with worse access to basic services have worse wellbeing outcomes

On the basis of our regression findings, it is neither possible to accept nor reject this hypothesis. The

overall picture is very mixed. In some cases we observe associations between worse access (proxied by

distance in minutes to a service – see Section 2.3.2 for a discussion on the choice of proxy) and better

wellbeing outcomes: for example, in Nepal a longer travel time to the health clinic is associated with

marginally greater levels of asset ownership, while in Pakistan the same independent variable is

associated with marginally lower levels of food insecurity. But in other cases we observe the reverse: in

both DRC and Uganda a longer travel time to water points is associated with lower levels of asset

ownership; in Nepal a longer travel time to the health clinic is associated with higher levels of food

insecurity; and in Sri Lanka a longer travel time to water points is associated with higher levels of food

insecurity. For most variables in this category, however, there does not appear to be a relationship of

any kind. Thus, with the possible exception of distance to water points (which is, in some cases,

positively related to both lower asset ownership and greater food insecurity), there does not seem to be

any clear association between access to services and these particular wellbeing outcomes.

Hypothesis 6: Households with better access to social protection and/or livelihood assistance have

better wellbeing outcomes

We also wanted to know whether households with better access to social protection and/or livelihood

assistance have better wellbeing outcomes. We find, in fact, that receipt of social protection and receipt

of livelihood assistance have different associations with wellbeing outcomes. In terms of social

protection, we asked households about receipt of the most common social protection transfers – such

as the Samurdhi cash transfer in Sri Lanka, to give just one example (see Annex 2, Table 21 for a list of

the transfers included in these variables in each country). Where it is statistically significant, receipt of

social protection tends to be associated with worse outcomes: in Sri Lanka, access to social protection

is associated with lower levels of asset ownership and in Nepal, Pakistan and Sri Lanka it is associated

with higher levels of food insecurity. The one opposite relationship here is that asset ownership in Nepal

is positively linked with receipt of social protection. Of course, this does not necessarily mean these

transfers make households worse off, but instead could be an indication of targeting (more on this

below).

Livelihood assistance was again tailored to country context and included agricultural support, e.g. seeds

and asset transfers, but also livelihood-specific training and other livelihood assistance (see Annex 2,

Table 22 for a list of the transfers included in these variables in each country). Receipt of livelihood

assistance is consistently associated with better wellbeing outcomes. In every country, we find that

households receiving livelihood assistance are likely to be better off in terms of asset ownership.

Results are less consistent when looking at food insecurity outcomes: only two countries exhibit a

statistically significant relationship (Pakistan, Uganda), but in both cases receipt of livelihood assistance

29

is associated with lower levels of household food insecurity. It seems quite clear, therefore, that while

receipt of social protection is associated with worse wellbeing outcomes, the opposite is true for

receipt of livelihood assistance.

4.3 What do these findings tell us about livelihoods and wellbeing in conflict-affected

places?

What can the SLRC baseline data tell us about what shapes livelihoods and wellbeing in conflict-

affected places? While there is a large literature on the determinants of livelihoods and wellbeing in

developing countries, far less focuses on conflict-affected places, or looks specifically at the role of

conflict.14 The analysis in this study allows us to make a further contribution to this question. Drawing

on the results of cross-country, comparable regressions, we can draw five key findings from the

baseline analyses.

First, the findings point to the importance of education in making a living: households with a higher

average level of education15 consistently have better wellbeing outcomes in terms of both proxy

indicators (household wealth and food insecurity). In other words, the more highly educated the head,

the wealthier and less food insecure the household. This is of concern, as it may mean that poorly

educated households do not have the education levels necessary to improve their wealth and food

security. Findings from some of the country baseline reports suggest that this trend does not look likely

to be reversed any time soon. For instance, in Uganda current primary education graduation rates in

northern Uganda are at 47%, and access to secondary school in northern Uganda is as low as 15% in

the villages (Mazurana et al., 2014).

In the analyses we explicitly examined the role of conflict in determining people’s wellbeing, looking at

exposure to fighting in the past three years, perceived safety in and around the community, and past or

current displacement. We find little evidence of displacement and conflict affecting wellbeing

outcomes (with the exception of DRC), and inconsistent findings on the relationship with perceived

safety. In other words, within our sample populations, recent experiences of conflict does not seem to

matter as much as other factors in shaping wellbeing of households in the sampled areas. Given the

short (three-year) recall period these regression models may be missing some longer-term effects of

conflict which may emerge in the second wave of data analysis. The Uganda data have already shown

that experience of serious crimes does have some long-term effects on household’s wellbeing and

livelihoods (the Uganda survey was tailored to measure the lasting impact of crimes experienced during

the conflict which makes it more reasonable to infer causality) (Mazurana et al., 2014). Nevertheless,

these findings are forcing us to re-examine the ways in which we carry out SLRC research and to

consider how far we are guilty of ‘conflict exceptionalism’ – that is, explaining everything we find

through the lens of conflict, or failing to recognise the other factors and features that are driving change

in people’s livelihoods and wellbeing just as much as conflict or fragility.

Our third finding is closely related to the second. Generally speaking, households that have experienced

a greater number of crimes and shocks are also likely to have worse food insecurity. However, we do

not find a clear relationship between shocks or crimes and wealth. These findings on food insecurity are

broadly consistent with studies from developing yet largely stable countries (Skoufias, 2003, reviews

some studies linking environmental shocks and food insecurity; see Yamano et al., 2003, and Dercon et

al., 2005, for Ethiopian examples; see Van den Berg, 2009, on Nicaragua; see Hoddinott, 2006, on

Zimbabwe). What is important is that our findings demonstrate that similar effects of shocks can be

found in conflict-affected places. Again, these findings reinforce the idea that it is not all about conflict:

households living in conflict-affected places face a wide range of risks and vulnerabilities – not just

14 A notable exception is the Households in Conflict Network: www.hicn.org. 15 See previous footnote on the measurement of education in the regressions.

30

those associated with violence and conflict – and these appear to affect people’s wellbeing just as

much as conflict and violence.

Fourth, on the whole, there is no consistent relationship between access to basic services and

wellbeing outcomes. While better access to water services does seem to have a reliably positive

association with wellbeing outcomes, neither access to health nor education has a consistent and

statistically significant link. This is somewhat surprising, since we might expect that better access to

health or education services would improve wellbeing outcomes in the medium to long run (indirectly,

through enhancing human capital). Given that we are looking only at a cross-section here, one

explanation might be that wealthier households can afford to live nearer, and choose to live nearer, to

these services. Another possible explanation could relate to the nature of our proxy indicators for

access to services. Alternatively – or additionally – it could plausibly be the case that it is the quality of

services that matters rather than access in a rudimentary sense (as research from Sierra Leone

suggests (Sacks and Larizza, 2012). These issues should be explored through further research.

Finally, we examined whether households with better access to social protection and/or livelihood

assistance are also more likely to have better wellbeing outcomes. In the first instance, our findings

look inconsistent: receipt of social protection tends to be associated with worse wellbeing outcomes.

However, rather than showing that social protection transfers are making households worse off,

causality may actually be the other way round. Many of the social protection transfers received by

beneficiary households are targeted at poor or vulnerable households (for example food aid). Hence,

these findings appear to suggest that social protection is well-targeted, but at this stage of the analysis

we cannot disentangle causality. For livelihood assistance, we are less sure about success of targeting.

Here we see a positive correlation with wellbeing outcomes. This suggests two possibilities: livelihood

assistance is targeted at viable households that already own some assets (tools, land etc.) in order to

increase effectiveness; or alternatively it is intended for poorer households but captured by elites. As

the baseline data is only based on one wave and we are looking at a whole range of different

interventions, we cannot say which of these more likely. We may be able to explore this issue further

using longitudinal data after the second wave and also in ongoing qualitative research. For example, in

Uganda, the SLRC team is using qualitative methods to follow up on survey respondents and attempt to

understand what has enabled some households to be more successful than others in improving their

livelihoods and wellbeing.

31

5 Which factors influence people’s access to

and experience of services, social

protection and livelihood assistance?

This section summarises and synthesises the analyses performed to answer two research questions:

Which factors influence access to basic services, social protection transfers and livelihood

assistance?

Which factors influence experience of basic services, social protection transfers and

livelihood assistance?

More specifically, we return to the original hypotheses presented in Box 2 in Section 2.3.2, drawing on

the cross-country findings to investigate the validity of each hypothesis in turn. For the comparative

analysis, we have focused on the regression findings from ten sets of regressions: five measuring the

determinants of access to health, education and water services, social protection and livelihood

assistance and another five measuring the determinants of the experience of these services or

transfers.

We measure access to services in terms of distance in minutes to the closest service provider last used

(for health, education and water) and for social protection and livelihood assistance access is defined

as someone in the household having received any social protection transfer or livelihood assistance in

the past year (12 months).

Experience is based on a range of perceptions question asked to respondents. For health and

education, experience is measured in terms of overall satisfaction with the service provided on a scale

of 1-5, for water, if clean water is being provided, and for social protection and livelihood assistance in

terms of the perceived impact of the transfer (see Section 2.3.2 for a detailed explanation). We have

structured and phrased these questions so that these subjective assessments of satisfaction are based

on actual experiences with the (service) provider, rather than general perceptions. The independent

variables for these analyses are set out in the analytical framework in Section 2.3.2.

This section is split into three parts. The first provides some basic descriptive statistics from each

country in order to illustrate key trends of access to and experiences of basic services, social protection

and livelihood assistance of our sample populations within each country. The second part draws on

regression findings from across the country studies to assess whether the hypotheses hold true. The

third part discusses the findings in a more narrative-based way.

5.1 Overview of access to and experiences with services, social protection and livelihood

assistance

Before delving into the regression findings, we give an overview of some of the key trends of access to

and experiences of basic services, social protection and livelihood assistance using cross-country

descriptive statistics.

Figures 10-12 show access to basic services in terms of our proxy indicator, distance travelled in

minutes. This indicator is interpreted as follows: the shorter the travel distance, the better access to the

services. As the figures show, on the whole, access to basic services is high, with consistently better

access in the three Asian countries and the lowest levels of access in the Uganda sample population.

Access to social protection and livelihood assistance is measured in terms of someone in the household

receiving a social protection transfer and/or livelihood assistance. The majority of households in our

sample populations do not receive social protection and livelihood assistance. Coverage was highest in

32

Nepal (38%) and lowest in Uganda (4%). Much of this coverage is accounted for by large-scale

programmes, although this is not always the case. For example, while 21% of households in the

Pakistan sample received the Benazir Income Support Programme (BISP) cash transfer, in DRC and

Uganda a large proportion of transfers were one-off forms of assistance from humanitarian agencies or

NGOs. In Uganda, for example, one-third of support received was a one-off transfer (Mazurana et al.,

2014). Access to livelihood assistance is ranges from between 15 and 16% of households in DRC,

Nepal and Uganda to 32% in Sri Lanka. Seeds and tools are the most frequently form of assistance

received across all the countries, with as many as 12% of households in Pakistan receiving such a

transfer (see Table 21 in Annex 2 for a breakdown of individual transfer coverage in each country

sample).

Figures 14-15 show respondents satisfaction with basic services, livelihood assistance and social

protection. It is apparent that satisfaction with basic services is high. More than 57% of respondents are

satisfied or very satisfied with the health services provided and satisfaction is consistency high across

the countries with the exception of Uganda. With regard to primary schools, we see even higher levels of

satisfaction, with 63% of respondents being satisfied or very satisfied, with respondents in the Uganda

and DRC sample having lower satisfaction levels. Satisfaction with water, using perception of water

quality as a proxy, is also high, again somewhat lower in DRC and Uganda. However, this is a cross-

sectional dataset, so the strikingly high satisfaction levels do not tell us much unless we put them into a

wider or longitudinal context. What we could be seeing is high satisfaction as a result of services

improving (perhaps even only marginally) compared to periods of conflict. With the second round of the

panel completed we will be able to identify individuals who changed their perceptions over time and

determine which other variables might be responsible. We will also be able to qualify our observation

that population average levels of satisfaction are high by comparing their level in the next time period.

The proportions of respondents perceiving that social protection and livelihood assistance had positive

impacts are lower, but still fairly high. On average, 64% of respondents perceive that livelihood

assistance has had a positive impact, ranging from 43% of respondents in Pakistan to 85% of

respondents in Nepal. Between 60% and 66% of respondents perceive that social protection has had a

positive impact. There is one major exception, however: only 21% of respondents in the Sri Lanka

sample perceived that social protection has had a positive impact. It is not clear why this is the case

and it may be linked to the sample of the study (see Mayadunne et al., 2014).

The next section focuses on regression analysis and we examine what the determinants of access to

and experiences of basic services, livelihood assistance and social protection are.

Figure 10: Distance to health clinic used (measured by time taken for a return journey), by country

4 3 1 4

30 14

9 8

12

25

31

22 23

16

26

52

66 68 69

19

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

30 minutes or

less

31-60 minutes

61-120 minutes

More than 2

hours

% taking time to

reach health

33

Figure 11: Distance to primary school used (measured by time taken for a return journey), by country

Figure 12: Distance to water source used, (measured by time taken for a return journey), by country

Figure 13: Households receiving social protection and livelihood assistance, by country

48

83 93

89

48

29

13

6 6

30

18

4 4

17

5 2

5

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

More than 2 hours

61-120 minutes

31-60 minutes

30 minutes or less

76

94 94 94

59

19

5 6 5

27

5 11

3

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

More than 2 hours

61-120 minutes

31-60 minutes

30 minutes or less

22

38

25 26

4

17 16

24

32

15

0

5

10

15

20

25

30

35

40

DRC Nepal Pakistan Sri Lanka Uganda

Social protection Livelihoods transfer

% receiving

transfer

% taking time to

reach primary

school

% taking time to

reach water source

34

Figure 14: Overall satisfaction with health facility used (% of respondents)

Figure 15: Overall satisfaction with primary school used

Figure 16: Perception of water quality, by country

3 11 13 15 5

19 40

22

38

21

7

0

58

36

62 57

60

21

3

11 13

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

Very satisfied

Satisfied

Neutral/Fairly satisfied

Dissatisfied

Very dissatisfied

% of

respondents

6 3 2

24

3

18 13

26

38

13

5

5 25

50

29

72 59

59

75

23

3 14 15

21

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan

girls

Pakistan

boys

Sri Lanka Uganda

Very satisfied

Satisfied

Neutral/Fairly

satisfied

Dissatisfied

Very dissatisfied

77 89 93 92

75

23 11 7 8

25

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

No

Yes

% of

respondents

perceiving

water as clean

and safe

% of

respondents

35

Figure 17: Transfer had any positive impact, by type of transfer and across countries

5.2 Examining the access to and experience of services/transfers hypotheses

We are interested here in what kinds of variables are associated with better or worse access to and

experience of basic services, social protection and livelihood assistance. There are a number of things

we might expect to be influential, from various household characteristics – education levels or wealth

levels, for example – to certain contextual features, such as (perceived) levels of safety in the local

environment and of course the way in which a service/transfer is implemented. In Section 2.3.1, we

outlined a set of five access to and experience of services/transfers hypotheses (Box 2). These

hypotheses contained statements regarding the kinds of relationships we would expect to see between

access/experience outcomes and a range of independent variables. In this section, we draw on findings

from country-level regression analysis to examine the validity of our hypotheses. We do so by looking

across a set of regression tables in an attempt to identify patterns and consistencies (see Annex 1 for

the full regression findings). In the third and final part of this section, we discuss what these findings tell

us about access to and experience of basic services, social protection and livelihood assistance in

conflict-affected places more generally.

Hypothesis 1: Wealthier and better-educated households have better access to basic services and

social protection and livelihood assistance

The regression findings do not fully confirm this hypothesis, providing only partial and weak evidence in

support of it. While households with a higher average education level on the whole have better access

to basic services, the relationship between wealth and access to services, social protection and

livelihood assistance is generally small and inconsistent.

The variable measuring the average education level of the adult household members is significant in

the regressions considering access to health, education, water and social protection (so not for

livelihood assistance), but never for any one sector in all five countries. In general, households with

better-educated adult members have better access to health, education and water services (i.e. shorter

distance), with the exception of water for DRC, and are less likely to receive social protection in Nepal

and Pakistan (but not so for Uganda). The education association is particularly pronounced in the case

of health; in Nepal for instance distance to the health centre is four minutes shorter for every one year

increase in the median years of education of the household members; in Uganda the time taken to

reach the health centre is more than 20 minutes shorter if the household head has a tertiary education.

The wealth of the household (as measured by the Morris index, an asset-based measurement) has a

much weaker link with access on the other hand, and, where significant, is often close to zero. More

wealthy households have better access to health and water for two of the countries, but we also find

62 62 66

21

60 62

85

43

77

53

0

10

20

30

40

50

60

70

80

90

DRC Nepal Pakistan Sri Lanka Uganda

SP

Livelihood

% of

respondents

who stated that

transfer had any

positive impact

36

worse access (i.e. longer distances to the health centre) for households in Nepal. This may be because

wealthier households in Nepal tend to use private health clinics, which are often located at a distance in

district capitals. In the case of social protection, we do indeed find that wealthier households in

Pakistan are more likely to access social protection, but the coefficient is small. The same holds for

access to livelihood assistance, where the Morris index only has a small but significant relationship for

Nepal and Sri Lanka.

Hypothesis 2: Households that are or have been displaced have worse access to and experiences of

basic services, social protection and livelihood assistance

For this hypothesis, we test whether current or past displacement affects access to and experiences of

basic services and social protection and livelihood assistance. On the whole, the regression findings do

not support this hypothesis. For the education, social protection and livelihood regressions the variable

is not significant in four countries. For access to health, the findings show that households that have

been displaced face worse access in Pakistan and Sri Lanka and better access (shorter distances) in

Nepal.16 In terms of satisfaction with health, having been displaced reduces the likelihood of being

satisfied for Uganda and Sri Lanka but has no statistically significant association in the other countries.

In terms of water access, having been displaced was associated with better access to water in DRC and

Uganda, but in Sri Lanka it was associated with slightly worse access. For satisfaction with water we find

similarly inconsistent patterns: having previously been displaced made it more likely that respondents

are satisfied in the case of Uganda, but the opposite holds for Nepal. In short, there are no clear or

consistent patterns emerging from the cross-country regressions, perhaps owing to the different

patterns of displacement and settlement of displaced persons (for instance, in DRC many displaced

persons end up in IDP camps, targeted by relief agencies who bring services directly to them, explaining

why distance to the water source is shorter).

In summary, the evidence is mixed and there is no consistent evidence showing that displaced

households have worse access to and experiences of basic services, social protection and livelihood

assistance. Instead we see that in some cases displaced households have better access to and

experiences of services, social protection and livelihood assistance. There is however some variation in

the way that displacement is measured across the five countries which could explain why the results

are inconsistent.

Hypothesis 3: Households that have recently experienced conflict or living in (perceived) unsafe

locations have worse access to and experiences of basic services and social protection and livelihood

assistance

The findings from regression analysis do not support this hypothesis and instead suggest that, to some

extent, households that have experienced conflict or feeling unsafe in the past three years have better

access to basic services, social protection and livelihood assistance (and to a lesser degree, better

experiences). For instance, households that have experienced fighting have better access to health for

Nepal and DRC, much better access to education for DRC, better access to water for Nepal and Uganda

and better access to social protection and livelihood services for DRC (and access was not significant in

all other cases). The conflict variable is insignificant in four countries on experience of the education

and health services and in three countries on impact of social protection,17 but relates positively and

significantly to satisfaction with water quality in DRC. The perceptions of safety variables do not have a

significant coefficient in any country for access to health, education, social protection and livelihood

assistance. Perceptions of safety are also not significant in explaining experience of the education

16 In Nepal the proxy used for displacement did not only capture displacement due to conflict but also voluntary mobility, for

example for family formation. This may explain why we find better access to services among those who have moved village in

Nepal since in some cases an element of choice was involved as to the destination. 17 For livelihood assistance impact the conflict variable was only included in one country regression but here also showed no

significant association.

37

service in four countries and are insignificant in explaining experience of the health service, water

quality, and social protection impacts in three countries. In the cases where perception of safety is

significant, it suggests that feeling safe is associated with better access – for instance, those who feel

safe have better access to water in Pakistan. There are no clear patterns for experience of services but

there are some trends within countries, for example in Nepal feeling safe is associated with better

perceptions of the health service and the impact of social protection and livelihood assistance. In

summary, perceptions of safety are associated with better access to basic services, social protection

and livelihood assistance in a limited number of cases – the perceptions of safety variable is significant

for at least one country for seven out of ten outcomes. Surprisingly, having experienced conflict in the

past three years is also associated with better access and, albeit to a lesser degree, better experiences,

but again this effect is only found in a limited number of regressions.

Hypothesis 4: Respondents that are satisfied with how a service/transfer is implemented are more

satisfied with the service/transfer overall

Here we are interested in how the specific experiences of using a service affect the overall satisfaction

with the service. In order to test this hypothesis, we considered a number of variables: people’s specific

personal experiences with the service (e.g. length of wait, timeliness of the transfer etc.), formal and

informal payments for the service/transfer, and who the service provider is. We consider this an

important relationship to examine, as it helps us determine whether the quality of services matters for

people’s satisfaction. Indeed, it is possible that satisfaction may be high even where quality is poor, as

past experiences and low expectations in the first place may shape people’s subsequent views quite

dramatically. Thus, there is merit in testing the link, even though it may seem initially obvious.

For all regressions that we ran on satisfaction with basic services and livelihood assistance, at least one

aspect of how services were implemented is significant and these mostly had a positive coefficient. In

other words, respondents being satisfied with the specific experience of health, education (e.g. number

of staff), and livelihood assistance (e.g. received on time) had a positive association with their overall

satisfaction with the service/transfer. Notable exceptions are satisfaction with aspects of education for

Nepal and timeliness of social protection receipt in Nepal and DRC. Nonetheless, on the whole we find

evidence in favour of the above hypothesis. For instance, concerning livelihood assistance, respondents

in Nepal, Sri Lanka and Uganda who reported receiving the transfer on time were more likely to perceive

that the transfer had a positive impact. In this context a positive impact is measured using the proxy of

whether the livelihood assistance improved agricultural production or improved another livelihood

activity. The variable measuring paying formal or informal fees is significant in two countries for

experience of health, significant in one country for experience of education, and significant in four

countries for experience of water. For all but two of the significant associations, paying fees decreases

the likelihood of satisfaction with the service. The exceptions are that paying formal fees increases the

likelihood of satisfaction with health for DRC and Nepal (informal and formal payments), and paying

official fees increases satisfaction with water in Uganda. In Pakistan, paying official fees for the school

that boys attend decreases satisfaction, while informal fees for the school that girls attend increases

the likelihood of satisfaction. The provider of the service has a significant association with overall

satisfaction with the service/transfer in two countries for education, health, and livelihood assistance,

and in three countries for water (and is significant in the one country in which it was included for

experience of social protection). Respondents are more likely to be satisfied with water when they

perceived it as being run by the government, an NGO or the community (as opposed to privately or by

nobody). Regarding livelihood assistance, in Pakistan respondents were more likely to report an impact

if the provider was perceived to be not the government and in Uganda respondents were slightly more

likely to report an impact if the provider was perceived to be an NGO. This shows that who implements

the service, in some cases affects overall satisfaction with the service. In summary, we have partial

evidence for this hypothesis, suggesting that, to some extent, people’s specific personal experiences

38

with basic services, social protection and livelihood assistance does affect their overall satisfaction

with the service/transfer.

Hypothesis 5: Households with worse access to basic services are less satisfied with these

services

Finally, we wanted to know if respondents with worse access to basic services (as proxied by distance)

are less satisfied with these services. We tested this hypothesis in the regressions on satisfaction with

health and education. None of the regressions provide evidence to support this hypothesis (i.e. distance

is always insignificant). In other words, having access to the service does not appear to influence

satisfaction with health and education services. This echoes the finding for Hypothesis 4, which shows

that the way a service is implemented and run matters to people’s perceptions of services, not the

service itself.

5.3 What do these findings tell us about access to and experience of basic services, social

protection and livelihood assistance in conflict-affected places?

What can the SLRC baseline data tell us about what determines access to and experience of basic

services, social protection and livelihood assistance in conflict-affected places? We have analysed this

question using cross-country, comparable regressions and specifically through testing the five

hypotheses that were drawn from our analytical framework. On the whole, the empirical findings only

partially support the proposed hypotheses: many of our independent variables are insignificant and

inconsistent, suggesting that the determinants of access and experiences are complex and to a large

extent context-specific. Nevertheless, we can draw four key findings from our baseline analyses.

Our findings on the role of household wealth in terms of access to services and experiences with

services are complex and are not always consistent across countries. In many cases the wealth of a

household does not seem to matter at all, with poorer and richer households seemingly having similar

access to services. This may be because of the generally high access to basic services (as shown in the

first section of this section) or it could be a weakness of our proxy, which only measures distance but

not other indicators of access. In other cases patterns are puzzling – for example that richer households

in Nepal have worse access (that is they face longer travel distances). Rather than showing worse

access, what this finding is probably telling us is that rich and poor households can access services in

different ways: poorer households in our sampled areas in Nepal tend to use close-by government

schools, whereas richer households tend to use more distant private schools. Also puzzling in first

instance is that richer households are more likely to access livelihoods services in Nepal and Sri Lanka.

While this seemingly looks like elite capture, these households are probably more likely to own land on

which agricultural inputs can be used, suggesting effective targeting if assistance is targeted at viable

smallholders. However, as we explained in the previous section, at this stage of the analysis we are

unable to disentangle causality. What we can conclude from this is that wealth may not be as

important in accessing services and determining perceptions as we expected. Where it is statistically

significant, it appears to affect access and experience in different ways.

We started this research expecting to find that households that are or have been displaced have worse

access to and experiences of basic services and social protection and livelihood assistance. As was

shown above, our findings based on our baseline data do not fully support this hypothesis. For some

services, in a few countries we do find worse access to services, but in other cases we find that

displaced households have better access to services and are more satisfied with services, notably in

the DRC. One of the reasons why we may be seeing these surprising patterns in DRC (and to a lesser

extent in Uganda) is displacement has often resulted in relocation to IDP camps where households are

then provided with basic services by humanitarian agencies. The implication is that access measured

through our proxy variable – distance – is relatively good. While we do not fully understand these

39

findings yet and should explore these relationships further using qualitative research, they show that we

should be questioning our implicit assumptions on the effects of displacement on access to services.

Although there is no consistent set of variables explaining why some respondents are more satisfied

with services than others there is some indication that people’s specific personal experiences with the

service heavily influences their overall level of satisfaction with basic services and livelihood

assistance (we do not find this for social protection). With very minor exceptions, we found that

respondents being satisfied with the specific experience had a positive association with their overall

satisfaction with the service/transfer. Regression analysis of respondents’ experience with basic

services and livelihood assistance suggests that factors such as ‘satisfaction with the availability of

medicine’, ‘satisfaction with the waiting time in the clinic’, ‘satisfaction with the number of teachers’

and ‘satisfaction with the quality of the teaching staff’ are strongly and positively associated with higher

levels of overall satisfaction with those services/assistance. That said, we do not observe these

relationships across all services for all countries, suggesting people may attach different levels of

importance to particular characteristics of different services. Personal characteristics of the respondent

appear to matter less; for example, the gender of the respondent is insignificant in all countries for

three of the satisfaction with service/transfer regressions and has inconsistent associations in the other

two (health and education), but only for Nepal and Pakistan. However, these findings do point to the

importance of specific experiences in shaping overall satisfaction with a service/transfer.

Two sets of findings in particular are surprising: first, that levels of satisfaction with basic services were

generally quite high, while access was not consistently high for all services or across all countries (see

Section 5.1); and second, that there seems to be no apparent link between better or worse access to

basic services and satisfaction with these services (Hypothesis 5). There are three possible

explanations for this finding. The first one is simple: respondents may have responded more positively

in order to please the interviewer – this is known as the ‘social desirability response bias’. We hope to

have prevented such a bias through careful introductions to the survey and careful phrasing and

ordering of the questions. Second, we may be seeing greater satisfaction as a result of post-conflict

recovery: even though services are patchy and of low quality, they may still be better than the services

provided before (which may well have been no services at all). We are not in a position to assess access

to services prior to the baseline, but, when we have collected longitudinal data for the second round of

the survey, we will be able to fully explore the relevance of changing levels of access on satisfaction

with services.

Finally, what may also explain these findings is that levels of access are still generally low in our sample

populations: in other words, exposure to (high or low quality) services is low and people find it difficult to

judge their quality, both because of low expectations and lack of experience in using these services. We

could be seeing information asymmetries – for instance respondents may be very satisfied with latrine

toilets because they have not experienced a flush toilet. The next round of the survey should provide

further clues to the relative roles of low expectations, lack of experience and changing levels of service

provision in explaining satisfaction with basic services).

40

6 Which factors explain people’s

perceptions of the government?

In this final analytical section, we summarise and synthesise the analyses performed to answer the

research question: ‘which factors influence people’s perception of government actors?’ More

specifically, we return to the original hypotheses presented in Box 3 in Section 2.3.3, drawing on the

cross-country findings to investigate the validity of each hypothesis in turn.

In some countries, our survey instruments included questions about people's attitudes towards informal

state or non-state actors, such as local kings (mwami) in DRC. However, for comparative purposes we

focus here on people's views of local and central government. Our surveys primarily measured trust and

confidence in government actors through perception-based questions. As such, responses to the

following two generic questions formed the dependent variables used in the country-level regression

analyses (the findings of which we now draw on):

To what extent do you feel the decisions of those in power in the local/central government

reflect your own priorities? [Respondents were asked to select a single response from five

possible responses, which ranged from 'Never / not at all' to 'Completely']

Do you agree with the following statement: the local/central government cares about my

opinions. [Respondents were asked to select either 'Yes', 'No' or 'Don't know']

This means that we have four separate sets of regression analyses to draw upon in this section (that is,

the above questions were asked in relation to both local and central government). While our regression

models were built around the perceptions data, our surveys also generated other types of evidence on

people's relationships with the formal state in each of the focus countries. For example, measures of

civic participation – such as participation in public meetings regarding service delivery – constitute

another way of investigating what is often termed state–society relations. Although we do not report

here on the factors associated with higher or lower levels of civic participation, we do examine whether

interactions of this nature help us explain variations in people's attitudes towards the government.

As with the previous ones, this section is split into three parts. The first provides some basic descriptive

statistics from each country in order to illustrate some key 'perceptions of government characteristics'

of our sample populations within each country. More specifically, we report on levels of civic

participation, presence and use of grievance mechanisms within service delivery mechanisms, and

perceptions of the government. We then draw on regression findings from across the country studies to

answer whether our hypotheses hold true. The third part discusses the findings in a more narrative way.

6.1 Overview of perceptions of government

We now describe perceptions of government across the five countries. Figure 18 shows responses to

the two research questions for local and central government. As outlined in Section 2.3.3, much

thought was given as to how levels of government should be defined and perceptions measured. Two

key findings can be drawn from the responses to the two questions.

First, respondents have overwhelmingly negative perceptions of both levels of government. For

instance, for central government, only between 4% (Pakistan) and 44% (Sri Lanka) of respondents

perceived that the central government cares about their opinion. With the exceptions of Sri Lanka and

Uganda, at least two-thirds of respondents feel that the priorities of local government never or almost

never reflect their own. For some countries, more than 90% of respondents feel this way. Whilst it is

important to not read too much into the absolute figures (and much more important to see how they

change over time), with the possible exception of Sri Lanka, these responses do paint a rather bleak

picture of people’s perceptions of government. In the next section we will look at the determinants of

41

people’s perception of government – this should provide some tentative insights into why this is the

case.

Second, on the whole, perceptions of central government are worse than for local government, and this

difference is statistically significant in all countries. The responses to the ‘local/central government

cares about my opinion’ question are more negative for central government in four of the countries. The

exception is DRC, where the reverse is true, though the difference is quite small and not statistically

significant. A greater share of respondents found that central government’s priorities never reflect their

own, than for local government, for Nepal, Sri Lanka, Pakistan and Uganda. For DRC, many more

respondents (75%) never or almost never felt that local government reflected their priorities than the

equivalent for central government (53%). Future qualitative fieldwork will dig deeper into the question of

why this is the case.

Figure 18: Comparisons of perceptions of local and central government

Local government cares about my opinion

Central government cares about my opinion

Statistically significant difference between local and central government for Nepal, Pakistan, Sri Lanka

and Uganda (1% level of significance, with exception of Pakistan where 5%)

86

66

94

40

60

14

34

6

60

40

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

Yes

No

86 79

96

56 64

14 21

4

44 36

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

Yes

No

% of

respondents

% of

respondents

42

Local government decisions reflect my priorities

Central government decisions reflect my priorities

Statistically significant difference between local and central government for DRC, Nepal, Pakistan, Sri

Lanka and Uganda (1% level of significance).

6.2 Examining the government hypotheses

We are interested here in what kinds of variables are associated with better or worse perceptions of

government responsiveness and legitimacy. There are a number of things we might expect to be

influential, from various individual- and household-level characteristics – gender and education levels,

for example – to certain contextual features, such as (perceived) levels of safety in the local

environment. Given that the existing literature suggests close links between access to or experience of

public services and state legitimacy, we might also expect service-related factors to influence people's

perceptions of the formal state. In Section 2.3.3 we outlined a set of six government hypotheses (Box

3). These hypotheses contained statements regarding the kinds of relationships we would expect to see

75

57

90

10

41

0

10

4

13

19

16

30

4

41

27

9 10

2

19

10 17

3

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

Always/ completely

To a large extent

In some areas

Almost never

Never

53

70

95

13

47

28

11

2

18

18

12 18

2

52

22

8 2 1

11 9

5 3

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

DRC Nepal Pakistan Sri Lanka Uganda

Always/ completely

To a large extent

In some areas

Almost never

Never

43

between people's perceptions of the government and a range of independent variables. In this section,

we draw on findings from country-level regression analysis to examine the validity of our hypotheses.

We do so by looking across a set of regression tables in an attempt to identify patterns and

consistencies (see Annex x). In the third and final part of this section, we discuss what these findings

tell us about state–society relations in conflict-affected places more generally.

Hypothesis 1: Gender and education level of respondents shape perceptions of government

From our regression analyses, we find very little evidence that either of these individual-level variables

– gender and education level of respondent – explain variations in people's perceptions of either the

local or central government. Across all country regressions, there are very few statistically significant

associations between these independent variables and the outcome variables. In Uganda, there is

evidence to suggest that female respondents are more likely to hold worse perceptions of both the local

and central government compared to male respondents – although the regression results were not

always statistically significant. Similarly, there is some evidence from Nepal to suggest that better-

educated respondents are likely to hold more positive perceptions of local and central government, but

again the story is inconsistent. Overall, we find (1) very little consistency within countries vis-à-vis the

links between these particular individual-level variables and perceptions of the government; and (2)

even less consistency across countries on the same links.

Hypothesis 2: Respondents living in households that have recently experienced conflict, who are living

in (perceived) unsafe locations, or who have recently experienced a shock/crime have worse

perceptions of the government

There are no clear patterns when we consider the relationships between these independent variables –

which measure in different ways levels of local insecurity – and perceptions of government. For every

outcome, we find that at least one of these independent variables of interest – perception of safety,

having experienced conflict in the last three years and having experienced shocks/crimes – has an

inconsistent association with the outcome, perception of government.

Let us first look at the influence of conflict. At the local government level, experiencing conflict in the

last three years does not seem to have much of an link with perceptions of government except in two

cases: in Nepal and DRC, there is some evidence to suggest that such experiences are associated with

worse perceptions of local government. There are even fewer clearer patterns at the central

government level, although some regression evidence from Uganda suggests that the experience of

conflict is associated with lower levels of trust amongst respondents. Furthermore, respondents who

experienced serious crimes in Uganda have worse perceptions of the government, regardless of who

committed the crimes (Mazurana et al., 2014).

We might expect that respondents who feel safe in their community or when moving around would have

greater trust in the government compared to those who feel unsafe. But regression results show a

mixed relationship: individuals who feel safe in Pakistan and Sri Lanka are likely to have greater trust in

the local government, while the opposite appears true for those in Uganda (and to an extent DRC and

Nepal). On the other hand, we find that individuals who report feeling safe are more likely to have a

positive perception of the central government in both Sri Lanka and Uganda (while in the other

countries there is no significant relationship). So, there is some limited evidence to suggest that greater

(perceived) local safety is associated with more positive perceptions of the government in Sri Lanka, but

in the other four countries there is either a mixed pattern or no statistically significant relationship at

all.

Our findings are perhaps most inconsistent when we consider the link between perceptions of

government and experience of shocks and crimes. While there is a higher number of statistically

significant associations with perceptions of the local/central government, there is great variation in the

direction of those relationships. In some cases, the regression results actually contradict each other.

44

For example, results from the logit regressions suggest that respondents from households experiencing

a higher number of shocks are less likely to trust the local government in Nepal. But multinomial

regression results suggest that these same individuals are more likely to 'largely' or 'completely' agree

that local government decisions reflect their priorities. There is a relatively consistent (and intuitive)

association is in Sri Lanka, where greater exposure to shocks or crimes is quite often associated with

worse perceptions of the government (at both the local and central government levels). The other

consistent relationship we find is that crimes have a negative influence on perceptions of local

government (for two countries – for the rest it is not significant). On the whole, however, there appears

to be a no linear relationship between the experience of shocks/crimes and people's attitudes towards

the government.

Hypothesis 3: Respondents living in households that have better access to basic services, social

protection, social protection or livelihood assistance have more positive perceptions of the

government

On the basis of our country-level regression analyses, there does not appear to be any particularly clear

or consistent relationship between access to services/transfers and perceptions of the government.

This is true at both the local and central government levels.

With the exception of a few cases of statistical significance, access to basic services (health and water)

is not generally associated with levels of trust in the government. Among the few exceptions, in Nepal

and Sri Lanka, those facing longer journey times to water sources are less likely to feel the local

government cares about their opinions.18

Although the story looks very similar for social protection and livelihood assistance, we do find some

semblance of a pattern in the results. In one of the multinomial regression outputs, three out of five

countries (DRC, Nepal, Sri Lanka) demonstrate a statistically significant relationship between receipt of

social protection and agreement with the statement that the central government’s decisions reflect

respondents’ priorities. The association does not hold at the local government level. In addition, there is

evidence from Sri Lanka that shows a positive association between receipt of livelihood assistance and

more positive perceptions of the government, at both the local and central government level. We do not,

however, see this in any other countries.

Hypothesis 4: Respondents who have a more positive experience with basic services have more

positive perceptions of the government

In examining this hypothesis, we looked at two things. First, we asked whether levels of satisfaction with

basic services (health and water) help explain variation in perceptions of the government. For health, we

have satisfaction data in both a general sense (‘overall, how satisfied are you with the health clinic?’) as

well as in a more specific sense (‘how satisfied are you with the availability of medicines / with the

number of staff at the facility / with waiting times, and so on?’). For water, we have self-reported

assessments of whether the water accessed is clean and safe. Second, we ask whether variation in

people’s perceptions of the government can be partly explained by the number of problems a

respondent has experienced with the basic services they use.

On the first, we find that there is no clear or consistent relationship between respondents’ satisfaction

with basic services and their perceptions of either the local or central government. This is true for both

their general levels of satisfaction and specific levels of satisfaction (see above). Although there are a

number of cases of statistical significance, the directions of coefficients do not always flow in an

intuitive way. For example, while respondents reporting satisfaction with health services in Sri Lanka are

more likely to think the local government cares about their opinions, respondents reporting the same in

18 Education variables were not included in the regression models, because households without children do not use schools.

Including the variables would mean reducing the number of observations in each country analysis.

45

Uganda are in fact less likely to think the local government cares about their opinions. There are also

some contradictory findings vis-à-vis people’s specific experiences with health services and perceptions

of government. In Nepal, for example, respondents who are dissatisfied with the number of staff at the

health facility are more likely to think that local government decisions ‘never’ or ‘almost never’ reflect

their priorities, while respondents who are satisfied with the availability of medicine are also more likely

to think that local government decisions ‘never’ or ‘almost never’ reflect their priorities. It is not clear

what might explain these inconsistencies. Again, the importance that different communities attach to

different elements or features of service delivery may play a role.

Our regressions also included the provider of the health facility and water source that the respondent

used as independent variables. There are significant associations between health provider and

perception of government in three countries, although in Nepal the association only exists for local

government while in Sri Lanka and Uganda the health provider influences both government levels.

Three countries also show significant associations between water quality and perception of government:

this only applies for local government in Pakistan but for both levels in Nepal and Uganda. In DRC there

were no significant associations for water or health provider. The regressions make it possible to see

the significance of different providers; for example, in Uganda we find that respondents more frequently

find that the local government’s decisions reflect their priorities if it is the government that provides

water. If the local health service is government-run, then respondents have more positive perceptions of

local government (in Nepal) and central government (in Sri Lanka and for one of the regressions in

Uganda).

On the second measure of experience – the number of problems reported with services – a much

clearer pattern emerges. We consistently find that the more problems experienced, the worse

perceptions a respondent holds of the government. This is particularly the case in Nepal, Sri Lanka and

Uganda, where logit regression results show that respondents experiencing a greater number of

problems with their services are less likely to agree that both the local and central government do not

care about their opinions. Moreover, in four out of five countries (Nepal, Pakistan, Sri Lanka, Uganda),

the higher the number of service-related problems experienced, the less likely a respondent is to feel

that local government decisions ‘very much’ or ‘always’ reflect their priorities. The same relationship

holds at the central government level in two countries (Nepal, Sri Lanka).

Hypothesis 5: Respondents who have access to grievance mechanisms within public services have

more positive perceptions of government

In order to test this hypothesis, we look at whether the presence of a grievance mechanism (or

complaints procedure) within different services explains variation in people’s perceptions of the

government. Within the country survey instruments, our questions were designed to ask respondents

whether they were aware of how to make a complaint. For this particular hypothesis, we are not

interested in whether grievance mechanisms have actually been used; we are concerned simply with

whether respondents know they exist.19

Although not uniform across all countries, in three countries we find a consistent relationship between

the presence of grievance mechanisms (so long as respondents are aware of them) and positive

perceptions of the government. In Nepal, Pakistan and Uganda, respondents who know of a way to

make a complaint are more likely to agree that the local government cares about their opinions. Results

from the local government multinomial regressions lend support to this relationship in Nepal and

Pakistan. Concerning the central government, we find exactly the same positive associations in the

same countries. Thus, while there is no evidence of a relationship in either DRC or Sri Lanka, there is

quite a consistent picture in three of our five countries.

19 This of course means that our variables are not an objective indicator of whether a grievance mechanism is actually in place.

46

Hypothesis 6: Respondents with higher levels of civic participation have more positive perceptions of

government

For this final hypothesis, we are interested in whether knowledge of or participation in community

meetings about service delivery is associated with perceptions of the government. We focus on two

independent variables: (1) whether a respondent is aware of such meetings and whether they attended

such meetings in the past year; and (2) whether a respondent had been consulted about services in the

past year.

Whilst not the case for every country, we find several statistically significant associations between

whether a respondent is aware of or participates in community meetings about services and their

perceptions of the government. At the local government level, respondents who either knew about or

attended such meetings in Nepal, Sri Lanka and Uganda were all more likely to agree that local

government cares about their opinion. Results from the multinomial regressions confirm this positive

relationship in Sri Lanka and Uganda. This variable also matters at the central government level in

Nepal and Sri Lanka (according to logit regression results) and in Uganda (according to multinomial

regression results).

There is perhaps an even clearer relationship between whether a respondent was consulted about

services and their views towards the government. Respondents who had been consulted in the past

year were more likely to agree that local government cares about their opinions in four countries (Nepal,

Pakistan, Sri Lanka, Uganda), and more likely to agree that central government cares about their

opinions in three countries (Nepal, Pakistan, Uganda). Multinomial regression results show a very

similar pattern.

Thus, on this final hypothesis, there is fairly clear evidence to suggest that higher levels of participation

vis-à-vis service delivery (or at least greater opportunities to participate) are associated with better

perceptions of the government.

6.3 What do these findings tell us about people’s perceptions of the government?

State legitimacy is a complex, difficult-to-measure concept. We have attempted to generate a series of

proxy indicators of state legitimacy by asking respondents in each of our countries’ sample populations

about their perceptions of government. The existing literature suggests that this is a viable approach to

the study of what is ultimately an intangible and fuzzy dimension of statehood (Carter, 2011; Herbert,

2013; Hilker and Kangas, 2011).

The kinds of questions we asked respondents generated information on government performance (as

measured by citizens’ attitudes) and people’s levels of trust and confidence in the government. In terms

of our analysis, we are most interested in what kinds of factors explain statistical variation in

perceptions – or, in other words, what factors appear to influence the way people think about their

government, and whether there are any patterns that emerge from one country to the next regarding

these influencing factors.

The first thing to say about our cross-country government data is that it is not directly comparable in a

strict sense. We cannot, for example, say with any validity that respondents in our Nepal sample

population have more favourable perceptions of the local government than respondents in our DRC

sample population: perceptions cannot be viewed in such an absolute and comparable way. However,

our data do enable us to identify and analyse common trends and patterns. It is quite clear, for

example, that respondents across all five countries generally think worse of the central level of

government relative to the local government (see Figure 18 above). This suggests a number of things.

First, from an analytical point of view, it appears to make sense to separate out different government

actors or layers of government; approaching the state or government as a single monolithic entity

conceals potential variations in perceived legitimacy from a citizen’s perspective. Second, there is

47

perhaps something to be said of proximity and visibility factors here. Conceptual work by SLRC, drawing

extensively on existing governance research by others, suggests that the ‘tangibility’ of government and

public service delivery matters for how people think and feel about the state (Wild et al., 2014). To put it

in fairly crude terms, it is thought that if the state is seen to be actively doing stuff – providing services,

creating jobs, addressing grievances, and so on – then legitimacy gains may follow. Evidence from the

Afrobarometer studies, for example, shows quite clearly that people in sub-Saharan Africa ‘judge the

quality of local government primarily in terms of whether they think elected leaders “deliver the goods”’

(Bratton, 2010: 1). Because local layers of government tend to be more proximal to processes of local

development, it perhaps follows that people are more likely to perceive them to be acting in their

interests. At the very least, local government organisations are in theory more closely connected to local

populations, even if they fail to provide much in the way of tangible development gains. It may be that

this idea of connection and geographical proximity has some inherent effect on the way in which people

think about the state; the greater physical and hierarchical distance between people and the central

layer of government potentially has a limiting effect on its capacity to build legitimacy. Or in other words,

when the ‘goods are delivered’, people are arguably more likely to attach a positive association to the

level of government most visible to them.

On a similar point, our survey data suggest that there is no straightforward relationship between access

to basic services and welfare transfers and people’s attitudes towards the government. One popular

way of conceptualising legitimacy is as a performance-based outcome. That is, states become

legitimate in the eyes of their citizens by making and meeting promises of social and material

improvement (Burnell, 2006). Things like service delivery, job creation and so on then take on an

instrumental dimension; while they have a certain intrinsic value, they are conceptualised here as a

means to separate ends (state legitimacy). Improvements in service delivery are thus framed as vital

components of state-building strategies – we might put this crudely as ‘buying legitimacy through giving

stuff’. By our measures of access to services and transfers, however, it is clear that simply ‘getting

something’ is not sufficient to ‘buy legitimacy’. We find no apparent relationship between people’s

access to health or water services and their views towards either local or central government. That is,

those with better access to these services do not tend to be any more likely to have better perceptions

of the government. When we look at access to social protection, there is slightly more evidence of a

positive relationship, but it does not hold for every country and is only applicable at the central

government level. In short, our survey data do not provide convincing evidence that people’s views of

government legitimacy can be robustly explained by the fact that they are simply receiving something.

If we push the analysis further, however, it becomes apparent that the quality of what people are getting

does in fact matter. Evidence from three countries shows that the more problems experienced with

services over the past year, the worse respondents generally thought about the government. This

suggests that people do care about whether their public services are functioning properly and as they

should; in other words, it is false to assume that the simple presence of a health clinic or primary school

is sufficient to substantially change the way people think and feel about the government. Our evidence

here is in line with findings from other countries. Of particular note is a study by Sacks and Larizza

(2012), which involved examining whether decentralisation of service delivery in post-conflict Sierra

Leone had any influence at all on citizens’ attitudes towards public authority. The authors conclude:

Local service provision plays a critical role in shaping citizens’ attitudes toward political authorities.

However, our results also suggest the devolution of power per se is not a sufficient condition for

building the trustworthiness of local authorities … Citizens’ trust and support is not unconditional. On

the contrary, bureaucratic honesty combined with the quality of local service provision is what really

matters to citizens. (Sacks and Larizza, 2012: 23; emphasis added).

However, we find what are probably the clearest relationships when we consider participation and

accountability characteristics of public services. There is fairly strong evidence from our surveys that the

inclusiveness of service delivery is often associated with how people view the government. For example,

48

in three countries we find that the existence of grievance mechanisms within public services –

essentially, complaints procedures – is significantly related to more positive perceptions of the

government. What’s more, our findings suggest that actually using the grievance mechanisms does not

appear to be a necessary condition for this relationship to hold; their simple presence seems to matter

in and of itself. On a very similar note, in every country apart from DRC we observe statistical

associations between levels of civic participation vis-à-vis service delivery and perceptions of the

government.20 More specifically, when respondents attended a community meeting about services (or

knew of such a meeting), or when they were consulted about services in their community, they were

more likely to think better of the government. Taken together, these findings suggest there is potentially

something about the way in which public services can act as a channel through which citizens and

public authorities interact (Van de Walle and Scott, 2011). Opening up this route – allowing citizens to

‘see’ their system of government at work in a tangible, everyday manner – appears to influence the way

people think about their state. In a sense, this is perhaps less about the services themselves, and more

about the kinds of mechanisms that promote engagement and exchange between citizens and formal

state. More broadly, our survey findings quite clearly suggest that investigating the detail of service

delivery – the specific ways in which services and transfers are designed and implemented – as

opposed to simply asking whether people have access to something, appears to be a fruitful line of

enquiry.

Beyond these key service-related findings, it has to be said that very little appears to clearly and

consistently shape people’s perceptions of the government. For example, we might expect there to be

variations in accordance with the gender of our respondents. Processes of state formation and state-

building are intimately connected to questions of power and its distribution – and there are, in turn,

striking gender dimensions to those questions. Research from five conflict-affected countries shows

that women were largely excluded from negotiations regarding the post-conflict political settlement, and

that the elites controlling those negotiations were often in fierce opposition to improving women’s rights

and political participation (Castillejo, 2011). In addition, a recent review of the links between gender

equality and state-building found that ‘the inclusion and presence of women and gender concerns

remains marginal and uneven at every stage [of the process]’ (Domingo et al., 2013: iv). Yet, we find no

consistent evidence that female respondents have systematically worse perceptions of the

government. If nothing else, this serves to emphasise the importance of looking at intersectional

aspects of personal identity as opposed to just simple, one-dimensional identifiers. We also find very

little evidence that exposure to conflict and shocks or experience of displacement significantly shape

attitudes towards the government, which is consistent with our findings in relation to livelihoods and

wellbeing outcomes. Similarly, how well-off a household is – in terms of wealth and food security – does

not appear to matter all that much; respondents’ perceptions of the government do not seem to be

dependent on material wellbeing at the household level.

20 It is important not to make the assumption that participation in community meetings in DRC necessarily indicate a lively and

active civil society. Instead, it must be recognised that since the Mobutu era, participation has been a critical way of getting

support (for example food aid) and attending meetings in the hope of receiving support has become embedded in regular

activities at community level.

49

7 Conclusions and initial policy implications

In 2012/13, the SLRC designed and implemented the first round of a panel survey in five conflict-

affected countries, generating cross-country data on livelihoods, access to and experience of basic

services, exposure to shocks and coping strategies, and people’s perceptions of government. This paper

synthesised the findings of the DRC Nepal, Pakistan, Sri Lanka and Uganda surveys, which was

delivered to a total of 9,769 households in September-October 2012 (for DRC, Nepal, Pakistan and Sri

Lanka) and in January 2013 (for Uganda). The sampling strategy combined purposive and random

sampling at different stages in order to select areas that are relevant to the main research questions

and of national relevance, while also being able to draw statistically significant conclusions at the study

and village level.

In order to address the broader research questions of the SLRC, we wanted some degree of

comparability. However, the fact that we have non-nationally representative surveys in completely

different settings means we cannot make direct comparisons. Thus, in our analysis we did not

aggregate the data and make direct comparisons across all countries, nor did we apply uniform

indicators or definitions across all countries – questions, indicators and variables were tailored to the

specific country contexts. For example, we were interested in the link between a household’s level of

wealth and their livelihoods, access to services and perceptions of government, and while we used the

same indicator (the Morris index) to measure wealth, the index was composed of different assets

across countries. What this survey has enabled us to do is identify some general trends and similarities

and differences between our case countries, keeping in mind that the findings are not fully

comparable.21

In the baseline surveys we considered four specific research questions (see Section 2.2) focusing on

the determinants of livelihoods and wellbeing, the determinants of basic services, social protection and

livelihood assistance access and experiences and, finally, the factors determining perceptions of local

and central government. Comparable analyses were run across countries for the country baseline

reports (de Milliano et al., 2015; Mayadunne et al., 2014; Mazarana et al., 2014; Shahbaz et al., 2014;

Upreti et al., 2014) and these were utilised in this synthesis report. We mainly drew on descriptive

statistics and regression analysis. The latter included OLS regressions, logit regression and multinomial

logit regressions.

Rather than giving a detailed summary of the report, in this conclusion we focus on six key findings that

have emerged from the synthesis of cross-country findings. These are findings that are particularly

interesting and surprising ‒ some challenge the hypotheses that we set out prior to conducting the

analysis and, to some extent, also challenge perceived wisdom on fragile and conflict-affected areas.

Where possible, we also draw relevant policy lessons.

First, we found that it is not all about conflict. While we find long-term effects of serious crimes in

Uganda, we find little evidence of conflict in the past three years, past or current displacement and

perceptions of safety clearly or consistently affecting wellbeing outcomes, access to services and

perceptions of the government in the other countries. In many of the sampled areas conflict is just one

risk amongst many. The lives of people in fragile and conflict-affected areas are not defined by conflict

and their choices and characteristics are not determined by fighting in the absolute sense. People in

these places often live in environments characterised by multiple vulnerabilities: conflict-related shocks

may be an important part of the landscape, but so too are ‘non-conflict related’ shocks, such as

environmental stresses, natural disasters, health shocks and economic shocks. These findings are

21 For example even if the same indicator was used, we did not directly compare numerical value across countries, nor did we

compare regression coefficients across countries.

50

forcing us to re-examine ‘conflict exceptionalism’: while we might intuitively expect conflict to

dramatically affect people’s livelihoods and wellbeing and perceptions of the government, our findings

are showing that other factors are often at least as important in driving outcomes.

There are some policy implications we can draw from this. Peoples’ lived experiences do not follow a

neat divide between conflict, post-conflict and development phases, but donor and aid programming

too often does – with abrupt shifts between different types of funding mechanisms and different types

of programming. For donors this implies that transitions between conflict and post-conflict approaches

and humanitarian and development funding need to be less abrupt and better layered. In practice this

means that policy makers in conflict-affected areas need to be concerned, for example, with health

shocks, floods and crop diseases, as well as with conflict-related displacement. It also highlights the

weakness of viewing states as either failed or not, or fragile or not, and the need for a more nuanced

typology of fragility that recognises that Uganda has more in common with Kenya or Tanzania than with

Afghanistan or the Solomon Islands.

Second, the findings point to the importance of education in making a living: households with more

highly educated household heads consistently have better wellbeing outcomes in terms of both proxy

indicators (household wealth and food insecurity). In other words, the more highly educated the head,

the wealthier and less food insecure the household. The literature already shows a strong correlation

between greater education and improved livelihoods, greater assets and other wellbeing outcomes in

developing countries more generally (Collinson, 2009; Ellis, 2003; Garrett and Ruel, 1999; Moser,

1998), and here we show that this pattern also holds in fragile and conflict-affected areas. From a

policy point of view this means that a stronger focus should be placed on secondary education. Our

findings suggest that primary school education makes a difference to wellbeing outcomes, but

recipients of secondary school education clearly have even higher wellbeing outcomes (caution should

be exercised in reading causality in this relationship).

Third, our findings on access to livelihood assistance suggest that, paradoxically, the transition from

conflict to post-conflict does not appear to result in renewed efforts to support livelihood rehabilitation

(with the caveat that at we are not yet able to measure change in access to livelihood assistance). At

the baseline, coverage of livelihood assistance reaches less than one-third of households – ranging

between 15% and 32% of sampled households across the case study countries. There is an apparent

gap in both effective strategies and effective programmes at sufficient levels of scale to support

processes of livelihoods recovery, provide social protection and stimulate employment and growth.

Further SLRC research is exploring whether, in post-conflict periods, donors are focusing on supporting

government systems to deliver health and education services at the expense of supporting livelihoods

and jobs.

Fourth, and perhaps one of the most unexpected findings, is that levels of satisfaction with basic

services are generally quite high across the countries (with some minor exceptions). Before trying to

understand this finding we would like to note that this does not mean households are in fact accessing

high-quality services – we are measuring self-reported satisfaction levels based on households’

experiences of the service. We highlight three possible explanations for this finding here, but there may

be others. One is respondents expressing positive perceptions owing to a social desirability bias. There

may also be actual post-conflict improvements in services and high levels of satisfaction due to

information asymmetries. On the one hand, we may be seeing greater satisfaction as a result of post-

conflict recovery: even though services are patchy and, possibly, of low quality, these services may be

better than the services provided previously (i.e. from a very low base). We are not in a position to

assess access to services prior to the baseline, but we will be able fully explore the relevance of

changing levels of access on satisfaction with services after the second round of the survey. What may

also explain these findings is that levels of access are generally low in our sample populations; in other

words, exposure to (high/low quality) services is low and people find it difficult to judge their quality,

both because of low expectations and lack of experience in using these services.

51

Fifth, with the exception of Sri Lanka, respondents have overwhelmingly negative perceptions of both

levels of government, particularly of central government. For instance, only between 4% (Pakistan) and

36% (Uganda) of respondents perceived that the central government cares about their opinion. With the

exceptions of Sri Lanka and Uganda, at least two-thirds of respondents feel that the priorities of local

government never or almost never reflect their own. For some countries, more than 90% of respondents

feel this way. These findings do not necessarily suggest that central government is doing a ‘worse job’.

Rather, there is perhaps something to be said of proximity and visibility factors here. Conceptual work

by SLRC, drawing extensively on existing government research by others, suggests that the ‘tangibility’

of government and public service delivery matters for how people think and feel about the state (Wild et

al., 2014). To put it in fairly crude terms, it is thought that if the state is seen to be actively doing stuff –

providing services, creating jobs, addressing grievances, and so on – then legitimacy gains may follow.

At the very least, local government organisations are in theory more closely connected to local

populations, even if they fail to provide much in the way of tangible development gains. It may be that

this idea of connection and geographical proximity has some inherent effect on the way in which people

think about the state; the greater physical and hierarchical distance between people and the central

layer of government potentially has a limiting effect on its capacity to build legitimacy. Or, in other

words, when the ‘goods are / are not delivered’, people are arguably more likely to attach a positive /

negative association to the level of government most visible to them.

Finally, one of strongest findings that emerged is that ‘it ain’t what you do, it’s how you do it’. We find

that access to basic services, social protection or livelihood assistance on their own determines neither

satisfaction with services nor perceptions of the government. Instead what we see is that satisfaction

with services is heavily dependent on how well that service is run. For example, respondents’

experience with basic services suggests that factors such as ‘satisfaction with the availability of

medicine’, ‘satisfaction with the waiting time in the clinic’, ‘satisfaction with the number of teachers’

and ‘satisfaction with the quality of the teaching staff’ are strongly and positively associated with higher

levels of overall satisfaction with those services. Quality, as opposed to simple presence, is the most

important factor. Furthermore, whether or not someone has access to a particular service, social

protection or livelihood assistance does not appear to matter much in explaining perceptions of the

government. Instead we see that the presence of grievance mechanisms and possibilities for

participation strongly influences perceptions of the government – even if these are not effective in

practice. What’s more, our findings suggest that actually using the grievance mechanisms does not

appear to be a necessary condition for this relationship to hold; their simple presence seems to matter

in and of itself. Taken together, these findings suggest there is potentially something about the way in

which public services can act as a channel through which citizens and public authorities interact (Van

de Walle and Scott, 2011). Opening up this route – allowing citizens to ‘see’ their system of government

at work in a tangible, everyday manner – appears to influence the way people think about their

government. In a sense, this is perhaps less about the services themselves, and more about the kinds

of mechanisms that promote engagement and exchange between citizens and government. Additional

evidence for this idea comes from a separate cross-country study into multi-stakeholder processes

carried out by the Peace, Security and Development Network (Stel et al., 2012). Multi-stakeholder

processes (MSPs) refer to initiatives which aim to bring together different stakeholders – including the

state, civil society, the private sector, beneficiary communities and international organisations – in

order to encourage collective action for service delivery. This study found that the impact of MSPs on

state legitimacy is ‘determined more by their throughput (the multi-stakeholder process) than by their

output (improvement of service delivery)’ (Stel et al., 2012: 12). Read alongside our baseline survey

findings, this research quite clearly suggests that investigating the detail of service delivery – the

specific ways in which services and transfers are designed and implemented – as opposed to simply

asking whether people have access to something, appears to be a fruitful line of enquiry. In other

words, a narrow focus on outputs – what gets delivered – is unlikely to provide us with many insights

into the ‘quality of the state’, as Fukuyama (2013) puts it.

52

Findings from our surveys find that there is no simple or linear relationship between access to services

and peoples’ perceptions of the legitimacy and performance of government. For donors this suggests a

need to be more cautious about the claims they make that supporting services contributes to state-

building. Having said that, it is still absolutely appropriate to support service delivery from a rights- and

needs-based perspective: services are critical to eliminating poverty, irrespective of whether they

contribute to government legitimacy. But if donors want to support state capacities to deliver services

then they may need to pay much more attention to how they are supporting services and more explicit

about how government capacities are being built at what level and be more modest about the likely

impact of this on people’s views about the performance and legitimacy of governments.

While some surprising and interesting findings have emerged from the analysis and synthesis of the

SLRC baseline surveys, we should keep in mind that these findings provide just a snapshot at one

moment in time. This data does not allow us to draw any conclusions on causality, nor does it allow us

to measure changes over time. In 2015 we will be going back to the same households and respondents

and interviewing them for wave 2 of our survey. Only then will we be able to measure changes in

livelihoods, access to services and perceptions of government over time and be able to make definitive

statements about factors that determine these. We will also be exploring some of the findings the SLRC

baselines have thrown up through in-depth qualitative research.

53

Box 5: Key findings from the South Sudan Survey

The survey for South Sudan is somewhat different in content and in objective from the other SLRC

surveys, and we have therefore not included its findings in this synthesis report. This box provides a

brief summary of the South Sudan baseline report (d’Errico et al., 2014), which focused on in Jonglei

and Upper Nile states.

Overall, the population in the sample is very young. There seems to be a ‘missing’ population cohort

of males in the 20-40-year-old group. It is not clear whether this is an impact of the war, the result of

labour out-migration from the survey area to seek economic opportunities in other parts of the

country, or simply a result of the timing of the survey. Education levels are very low, with only a small

proportion of the sample reporting having received any formal education.

Livelihoods in the survey area are severely constrained, with limited options for expansion or

diversification. Livestock – and specifically cattle – have been the traditional mainstay of livelihoods

in much of the survey area. However, a very high proportion (>50%) of the population now report

owning no cattle, and other livestock have not, for the most part, taken the place of cattle. While the

survey data do not address this question directly, there is evidence of substantial shifts in livestock

ownership in very recent times due to cattle raiding (at least in Jonglei), so until more detailed survey

information is available, it is difficult to determine exactly how livestock ownership has shifted and

how much this affects livelihood change. Apart from livestock, people face limited options. Well over

half the households surveyed reported having no cultivation in the year prior to the survey, and a

quarter reported cultivating less than one feddan. Those who report some agricultural activities

mostly reported having cultivated only one kind crop in the previous season.

For all of the above reasons, the survey results demonstrate very high levels of household food

insecurity. The survey included six different indicators of food insecurity, and while each tells a

slightly different story, the overall picture is one in which food insecurity is well above humanitarian

emergency thresholds. The nutritional status of children under five years of age corroborates this

finding. Limited ownership of livestock and other assets, limited engagement in cultivation, the lack

of alternative livelihood activities, the frequency of shocks, and the high prevalence of food insecurity

and undernutrition all point to a livelihoods crisis of major dimensions.

Overall views about the quality of governance and about the state were more positive than answers

about livelihoods, access to services, or participation in governance functions might suggest. More

than 70% of respondents felt that local government never or almost never cares about their opinions,

although a minority believed rather strongly that the government does care about their views – and it

should be noted that given the mostly mono-ethnic character of the survey sample, this response is

not related to a dominant or minority ethnic status. Given the extent to which governance questions

about South Sudan are often dominated by narratives about ethnicity, this finding is of note. Overall,

the survey data offer little in the way of a demonstrable link between access to services and

perceptions of the state. Access to services is not good, but this is not significantly correlated (with a

few exceptions) with perceptions of the state.

54

References

ADB (Asian Development Bank) (2010) ‘Overview of Gender Equality and Social Exclusion in Nepal’. Kathmandu:

ADB.

Adhikari, T.P., Thapa, F.B., Tamrakar, S., Magar, P.B., Hagen-Zanker, J. and Babajanian, B. (2014) How does social

protection contribute to social inclusion in Nepal? London: ODI

Adato, M. and Bassett, L. (2009) ‘Social protection to support vulnerable children and families: the potential of

cash transfers to protect education, health and nutrition’, Aids Care, 21(S1), 60-75.

Arceneaux, K. and Stein, R.M. (2006) ‘Who is held responsible when disaster strikes? The attribution of

responsibility for a natural disaster in an urban election’, Journal of Urban Affairs, 28(1), 43-53.

Annan, J., Blattman, C. and Horton, R. (2006) The state of youth and youth protection in northern Uganda. Uganda:

UNICEF.

Ananth Pur, K. and M. Moore (2009) ‘Ambiguous Institutions: Traditional Governance and Local Democracy in

Rural India’, Journal of Development Studies 46(4): 603-623.

Anderson, J.R. and Feder, G. (2004) ‘Agricultural extension: Good intentions and hard realities’, The World Bank

Research Observer, 19(1), 41-60.

Babajanian, B., Hagen-Zanker, J. and Holmes, R. (2014) How do social protection and labour programmes

contribute to social inclusion? Evidence from Afghanistan, Bangladesh, India and Nepal. ODI Report.

Badiuzzaman, M., Cameron, J. and Murshed, S.M. (2011) Household decision-making under threat of violence: A

micro level study in the Chittagong Hill Tracts of Bangladesh (No. 39). MICROCON-A Micro Level Analysis of Violent

Conflict.

Barakat, S. (2009) ‘The failed promise of multi-donor trust funds: aid financing as an impediment to effective

state-building in post-conflict contexts’, Policy studies, 30(2): 107-126.

Barnett, M. (2006) ‘Building a republican peace: Stabilizing states after war.’ International security, 30(4), 87-112.

Bastagli, Francesca (2014) Delivering social protection in the aftermath of a shock: Lessons from Bangladesh,

Kenya, Pakistan and Viet Nam. London: ODI.

Baulch, B. and McCulloch, N. (2002) ‘Being poor and becoming poor: Poverty status and poverty transitions in

rural Pakistan’, Journal of Asian and African Studies, 37(2): 168-185.

Beath, A., Christia, F. and Enikolopov, R. (2012) ‘Winning hearts and minds through development? Evidence from

a field experiment in Afghanistan’

(https://www.princeton.edu/~pcglobal/conferences/methods/papers/beath.pdf).

Bertrand, M., Mullainathan, S. and Miller, D. (2003) ‘Public policy and extended families: Evidence from pensions

in South Africa’, The world bank economic review, 17(1): 27-50.

Bianchi, M. and Bobba, M. (2010) Financial Constraints and Occupational Choices: Evidence from Rural Mexico.

Blattman, C., Fiala, N. and Martinez, S. (2013) Credit constraints, occupational choice, and the process of

development: long run evidence from cash transfers in Uganda. SSRN Scholarly Paper.

Block, S. and Webb, P. (2001) The dynamics of livelihood diversification in post-famine Ethiopia. Food policy,

26(4): 333-350.

Boulding, C.E. and Nelson-Núñez, J. (2014) ‘Civil Society and Support for the Political System in Times of

Crisis’, Latin American Research Review, 49(1): 128-154.

Branisa, B., Klasen, S. and Ziegler, M. (2013) ‘Gender Inequality in Social Institutions and Gendered Development

Outcomes’, World Development 45: 252–268.

55

Bratton, M. (2007) Are You Being Served? Popular Satisfaction with Health and Education Services in Africa.

Afrobarometer Working Paper 65.

Bratton, M. (2010) Citizen perceptions of local government responsiveness in Sub-Saharan Africa. Afrobarometer

Working Paper 119.

Bratton, M. (2012) Citizen perceptions of local government responsiveness in Sub-Saharan Africa. World

Development, 40(3), 516-527.

Brinkerhoff, D. W. (2005) ‘Rebuilding governance in failed states and post‐conflict societies: core concepts and

cross‐cutting themes’, Public Administration and Development, 25(1): 3-14.

Brück, T. (1997) Macroeconomic effects of the war in Mozambique. Queen Elizabeth House Working Paper 11.

University of Oxford.

Burnell, P. (2006) ‘Autocratic opening to democracy: why legitimacy matters’, Third World Quarterly 27(4): 545-

562.

Carpenter, S., Slater, R. and Mallett, R. (2012) Social protection and basic services in fragile and conflict-affected

situations. Working paper 8. London: Secure Livelihoods Research Consortium

Carr, E.R. (2013) ‘Livelihoods as Intimate Government: Reframing the logic of livelihoods for development’, Third

World Quarterly, 34(1): 77-108.

Carter, D. (2011) ‘Sources of state legitimacy in contemporary South Africa: a theory of political goods’. Working

paper 134. Afrobarometer.

Castillejo, C. (2011). Building a state that works for women: integrating gender into post-conflict state building.

FRIDE Working Paper No. 107. Madrid: Fundacio ́n para las Relaciones Internacionales y el Dia ́logo Exterior.

Christian, M., Safari, O., Ramazani, P., Burnham, G. and Glass, N. (2011) ‘Sexual and gender based violence

against men in the Democratic Republic of Congo: effects on survivors, their families and the community’,

Medicine, conflict and survival, 27(4), 227-246.

Christensen, T. and Lægreid, P. (2005) ‘Trust in government: The relative importance of service satisfaction,

political factors, and demography’, Public Performance & Management Review, 28(4): 487-511.

Cockcroft, A., Khan, A., Ansari, N.M., Omer, K., Hamel, C. and Andersson, N. (2011) ‘Does contracting of health

care in Afghanistan work? Public and service-users' perceptions and experience’, BMC health services research,

11(Suppl 2), S11.

CBS and NPCS (Central Bureau of Statistics and National Planning Commission Secretariat) (2012) National Living

Standard Survey 2011/12. Kathmandu: CBS and NPCS.

DFID (Department for International Development) (n.d) ‘Regional Dimensions of Poverty and Vulnerability in

Nepal’. Background Report. Kathmandu: DFID

DFID (2011) ‘Cash Transfers Literature Review’. London: Policy Division, DFID.

Della Porta, D. (2000) ‘Social Capital, Beliefs in Government, and Political Corruption’, in S.J. Pharr and R.D.

Putnam (eds) Disaffected Democracies: What's Troubling the Trilateral Countries? Princeton, NJ: Princeton

University Press.

Dercon, S., Hoddinott, J. and Woldehanna, T. (2005) ‘Shocks and consumption in 15 Ethiopian villages, 1999–

2004’, Journal of African Economies 14(4), 559-585.

Dercon, S., Gilligan, D.O., Hoddinott, J. and Woldehanna, T. (2009) ‘The impact of agricultural extension and roads

on poverty and consumption growth in fifteen Ethiopian village’, American Journal of Agricultural

Economics, 91(4), 1007-1021.

D’Errico Karolina Kozlowska and Daniel Maxwell (2014) ‘Surveying livelihoods, service delivery and governance:

baseline evidence from South Sudan’ (SLRC: London)

56

Domingo, P. et al. (2014) ‘Assessment of the evidence of links between gender equality, peacebuilding and

statebuilding: literature review’, (ODI: London)

Ellis, F. (2003) A livelihoods approach to migration and poverty reduction. Paper commissioned by the Department

for International Development (DFID: London).

Ellis, F. (1998) ‘Survey article: household strategies and rural livelihood diversification’. Journal of Development

Studies 35(1): 1-38.

Ellis, F. (2000) ‘The determinants of rural livelihood diversification in developing countries’. Journal of Agricultural

Economics 51(2): 289-302

ESCAP (2007) ‘Urban poverty and the working poor: facing the challenges of urbanization and urban poverty in

Asia and the Pacific. Committee on Poverty Reduction, 24 September (E/ESCAP/CPR(4)/4).

Espinal, R., Hartlyn, J. and Kelly, J.M. (2006) ‘Performance still matters explaining trust in government in the

Dominican Republic’, Comparative Political Studies, 39(2): 200-223.

European Report on Development (2009) Overcoming Fragility in Africa. Robert Schuman Centre for Advanced

Studies.

Fernando, P. and Moonesinghe, S. (2012) Livelihoods, basic services and social protection in Sri Lanka. Working

Paper 6. London: Secure Livelihoods Research Consortium.

Finkel, S.E., Sabatini, C.A. and Bevis, G.G. (2000) ‘Civic education, civil society, and political mistrust in a

developing democracy: the case of the Dominican Republic’, World Development 28(11): 1851-1874.

Fukuyama, F. (2013) ‘What is governance?’ Governance 26(3): 347-68.

Garrett, J.L. and Ruel, M.T. (1999) ‘Are determinants of rural and urban food security and nutritional status

different? Some insights from Mozambique’, World Development 27(11): 1955-1975.

Gates, S., Hegre, H., Nygard, H.V. and Strand, H. (2012) ‘Development Consequences of Armed Conflict’. World

Development 40(9): 1713-22.

De Haan, L. and Zoomers, A. (2005) ‘Exploring the frontier of livelihoods research’, Development and Change

36(1): 27-47.

Hagen-Zanker, J., Mallett, R., Ghimire, A., Shah, Q., Upreti, B. and Abbas, H. (2014) Migration from the margins:

mobility, vulnerability and inevitability in mid-western Nepal and north-western Pakistan. London: Secure

Livelihoods Research Consortium.

Hagen-Zanker, J. and Azzarri, C. (2010) ‘Are internal migrants in Albania leaving for the better?’ Eastern European

Economics 48(6): 57-84.

Herbert, S. (2013) ‘Perception surveys in fragile and conflict-affected states’. GSDRC Helpdesk Research Report.

Birmingham: GSDRC.

Hilker, L. and Kangas, A. (2011) ‘DFID’s use of surveys and polls in conflict, security and justice’. Unpublished

Draft Report.

Hoddinott, J. (2006) ‘Shocks and their consequences across and within households in rural Zimbabwe’, Journal of

Development Studies 42(2): 301–321.

Hossain, N. (2010) ‘School exclusion as social exclusion: the practices and effects of a conditional cash transfer

programme for the poor in Bangladesh’, Journal of Development Studies 46(7): 1264-1282.

Inglehart, R. and Welzel. C. (2005) Modernization, Cultural Change and Democracy: The Human Development

Sequence. New York: Cambridge University Press.

Kennedy, E. and Peters, P. (1992) ‘Household food security and child nutrition: the interaction of income and

gender of household head’, World Development 20(8): 1077-1085.

Khandker R.S., Pitt, M. and Fuwa, N. (2003) Subsidy to promote girls secondary education: The female stipend

programme in Bangladesh (http://mpra.ub.uni-muenchen.de/23688/1/MPRA_paper_23688.pdf).

57

Klingemann, H-D. (1999) ‘Mapping Political Support in the 1990s: A Global Analysis.’ In P. Norris (ed.) Critical

Citizens: Global Support for Democratic Governance. New York: Oxford University Press: 31-56.

Kirsch, T.D., Wadhwani, C., Sauer, L., Doocy, S. and Catlett, C. (2012) ‘Impact of the 2010 Pakistan floods on rural

and urban populations at six months’, PLoS currents 4.

Konte, M. (2014) Gender difference in support for democracy in Sub-Saharan Africa: Do social institutions

matter? UNU-MERIT Working Paper 2014-009.

Kruk, M.E., Freedman, L.P., Anglin, G.A. and Waldman, R.J. (2010) ‘Rebuilding health systems to improve health

and promote statebuilding in post-conflict countries: A theoretical framework and research agenda’, Social

Science & Medicine 70(1): 89-97.

Logan, C. and Bratton, M. (2006) The political gender gap in Africa: similar attitudes, different behaviours.

Afrobarometer Working Paper No. 58.

Lægreid, P. (1993) Opinion og offentlig sektor (Public opinion and the public sector), in P. Lægreid and J.P. Olsen

(eds): Organisering av offentlig sector (Organizing the public sector). Oslo: TANO.

Mallett, R. and Slater, R. (2012) Growth and livelihoods in fragile and conflict-affected situations. Working paper

9. London: Secure Livelihoods Research Consortium

Mamdani, M. and Bangser, M. (2004) ‘Poor people's experiences of health services in Tanzania: a literature

review’, Reproductive Health Matters 12(24): 138-153.

Maxwell, D. and Caldwell, R. (2008) ‘The Coping Strategies Index: Field Methods Manual’, Cooperative for

Assistance and Relief Everywhere, Inc. (CARE).

Mayadunne, G., Mallett, R. and Hagen-Zanker, J. (2014) Surveying livelihoods, service delivery and governance:

baseline evidence from Sri Lanka. London: Secure Livelihoods Research Consortium.

Mazurana, D., Marshak, A., Opio, J.H. and Gordon, R. (2014) Surveying livelihoods, service delivery and

governance: baseline evidence from Uganda. London: Secure Livelihoods Research Consortium.

McDonagh, J., Ellis, F. and Freeman, H.A. (2005) ‘Crop-based farming systems and diverse livelihoods in Uganda’.

In F. Ellis and H.A. Freeman (eds) Rural Livelihoods and Poverty Reduction Policies. London: Routledge: 215-234.

Meagher, K. (2005) ‘Social capital or analytical liability? Social networks and African informal economies’ Global

Networks, 5(3): 217-238.

Menkhaus, K. (2007) ‘Governance without government in Somalia: spoilers, state building, and the politics of

coping’, International Security 31(3): 74-106.

Mertens, T.E., Fernando, M.A., Marshall, T.D., Kirkwood, B.R., Cairncross, S. and Radalowicz, A. (1990)

‘Determinants of water quality, availability and use in Kurunegala, Sri Lanka’, Tropical Medicine and Parasitology

41(1): 89-97.

De Milliano, C.W.J., Ferf, A., Oude Groeniger, J. and Mashanda, M. (2015) Surveying livelihoods, service delivery

and governance: baseline evidence from DRC. London: Secure Livelihoods Research Consortium.

Mishler, W. and Rose, R. (2001) ‘Political Support for Incomplete Democracies: Realist vs. Idealist Theories and

Measures’, International Political Science Review 22(4): 303-320.

Morris, S.D. (1991) Corruption and Politics in Contemporary Mexico. Tuscaloosa: University of Alabama Press.

Morris, S., Carletto, C., Hoddinott, J. and Christiaensen, L.J.M. (1999) Validity of Rapid Estimates of Household

Wealth and Income for Health Surveys in Rural Africa. Food and Consumption Division Discussion Paper No. 72.

Moser, C. (1998) ‘Reassessing urban poverty reduction strategies: The asset vulnerability framework’, World

development 26(1): 1-19.

Mrisho, M., Obrist, B., Schellenberg, J.A., Haws, R.A., Mushi, A.K., Mshinda, H., Tanner, M. and Schellenberg, D.

(2009) ‘The use of antenatal and postnatal care: perspectives and experiences of women and health care

providers in rural southern Tanzania’, BMC pregnancy and childbirth 9(1): 10.

58

NPC (National Planning Commission) (2011) NLSS statistical report, volume II. Kathmandu: NPC

Obrist, B., Iteba, N., Lengeler, C., Makemba, A., Mshana, C., Nathan, R., Alba, S., Dillip, A., Hetzel, M.W.,

Mayumana, I., Schulze, A. and Mshinda, H. (2007) ‘Access to health care in contexts of livelihood insecurity: a

framework for analysis and action’, PLoS Medicine 4(10): e308.

OECD (2010) The State’s Legitimacy in Fragile Situations: Unpacking Complexity. Paris: OECD

OECD (2011) ‘Contracting out government functions and services in post-conflict and fragile states: examples

from the health sector in Latin America and the Caribbean’. Partnership for Democratic Governance. Workshop

discussion paper. Paris: OECD.

Osofian, W. (2011) ‘Towards strengthening state–citizen relationship in fragile states and environments: the role

of cash transfer programmes’. Conference on Social Protection for Social Justice, Brighton, 13-15 April 2011.

Owens, T., Hoddinott, J. and Kinsey, B. (2003) ‘The Impact of Agricultural Extension on Farm Production in

Resettlement Areas of Zimbabwe’, Economic Development and Cultural Change 51(2): 337-357.

Pain, A. and Lautze, S. (2002) Addressing livelihoods in Afghanistan (No. 14647). Kabul: Afghanistan Research

and Evaluation Unit.

Quisumbing, A.R. (2003) ‘Food aid and child nutrition in rural Ethiopia’, World Development 31(7): 1309-1324.

Ribas, R.P. (2013) ‘Direct and Indirect Effects of Cash Transfer on Entrepreneurship’. Available at SSRN 2302802.

Rockers, P.C., Kruk, M.E. and Laugesen, M.J. (2012) ‘Perceptions of the health system and public trust in

government in low-and middle-income countries: evidence from the World Health Surveys’, Journal of health

politics, policy and law 37(3): 405-437.

Rockmore, M.E. (2012) Three Essays On Conflict, Insecurity, And Livelihoods (Doctoral dissertation, Cornell

University).

Rose-Ackerman, S. (2008) ‘Corruption and government’, International Peacekeeping 15(3): 328-343.

Sabharwal, N.S., Mishra, V.K., Naik, A.K., Holmes, R. and Hagen-Zanker, J. (2014) How does social protection

contribute to social inclusion in India? Evidence from the National Health Insurance Programme (RSBY) in

Maharashtra and Uttar Pradesh. London: Overseas Development Institute.

Sacks, A. and Larizza, M. (2012) Why quality matters: rebuilding trustworthy local government in post-conflict

Sierra Leone. Policy Research Working Paper 6021. Washington, DC: World Bank.

Schilling, J., Opiyo, F.E. and Scheffran, J. (2012) ‘Raiding pastoral livelihoods: motives and effects of violent

conflict in north-western Kenya’, Pastoralism 2(1): 1-16.

Scoones, I. (1995) ‘Investigating difference: applications of wealth ranking and household survey approaches

among farming households in southern Zimbabwe’, Development and Change 26(1): 67-88.

Seligson, M.A. (2002) The impact of corruption on regime legitimacy: A comparative study of four Latin American

countries. The Journal of Politics 64(2): 408-433.

Shahbaz, B., Shah, Q.A., Suleri, A., Kazmi, M.A. and Commins, S. (2014) Surveying livelihoods, service delivery and

governance: Baseline evidence from Pakistan. London: Secure Livelihoods Research Consortium.

Shemyakina, O. (2011) ‘The effect of armed conflict on accumulation of schooling: Results from

Tajikistan’, Journal of Development Economics 95(2): 186-200.

Slater, R., Mallett, R. and Carpenter, S. (2012) Social protection and basic services in conflict-affected situations:

what do we know? SLRC Briefing Paper 2. London: Secure Livelihoods Research Consortium.

Skoufias, E. (2003) ‘Economic crises and natural disasters: Coping strategies and policy implications’, World

Development 31(7): 1087-1102.

SLRC (2015) “Panel survey process paper”

SLRC forthcoming “Gender synthesis”

59

Steimann, B. (2005) Livelihood strategies in North West Pakistan. IP-6 Working Paper 5. Zurich: NCCR, North-

South, Development Study Group, University of Zurich.

Tschirley, D.L. and Weber, M.T. (1994) ‘Food security strategies under extremely adverse conditions: The

determinants of household income and consumption in rural Mozambique’, World Development 22(2): 159-173.

Takasaki, Y., Barham, B.L. and Coomes, O.T. (2001) ‘Amazonian peasants, rain forest use, and income generation:

the role of wealth and geographical factors’, Society & Natural Resources 14(4): 291-308.

UNDP (UN Development Programme) (2009) Nepal Human Development Report 2009. Kathmandu: UNDP.

Upreti, B.R., Upreti, P., Hagen-Zanker, J., KC, S. and Mallett, R. (2014) Surveying livelihoods, service delivery and

governance: baseline evidence from Nepal. London: Secure Livelihoods Research Consortium.

Vallings, C. and Torres, M.M. (2005) Drivers of Fragility: What Makes States Fragile? Poverty Reduction in Difficult

Environments (PRDE) Working Papers series. London: Department for International Development.

Voors, M.J., Nillesen, E.E., Verwimp, P., Bulte, E.H., Lensink, R. and Van Soest, D.P. (2012) ‘Violent conflict and

behavior: a field experiment in Burundi’, The American Economic Review 102(2): 941-964.

Wafula, C.O., Kaseje, D.C.O., Ochieng, B.M. and Were, A.O.W. (2013) ‘Economic impact of HIV/AIDS on rural

households in Suba District, Kenya’, Developing Country Studies 3(3): 152-159.

De Waal, A. and Whiteside, A. (2003) ‘New variant famine: AIDS and food crisis in southern Africa’, The Lancet

362(9391): 1234-1237.

Van de Walle, S. and Scott, Z. (2011) ‘The Political Role of Service Delivery in State‐Building: Exploring the

Relevance of European History for Developing Countries’, Development Policy Review 29(1): 5-21.

Wang, X. and Hunter, P.R. (2010) ‘A systematic review and meta-analysis of the association between self-reported

diarrheal disease and distance from home to water source’, The American Journal of Tropical Medicine and

Hygiene 83(3): 582-584.

Weijs, B., Hilhorst, D.J. and Ferf, A. (2012) Livelihoods, basic services and social protection in Democratic

Republic of the Congo. London: Secure Livelihoods Research Consortium.

World Bank (2006) ‘Unequal Citizens. Gender, Caste and Ethnic Exclusion in Nepal. Summary’. Kathmandu: World

Bank Nepal and DFID Nepal.

Yamano, T., Alderman, H. and Christiaensen, L. (2005) ‘Child growth, shocks, and food aid in rural

Ethiopia’, American Journal of Agricultural Economics 87(2): 273-288

60

Annex 1: Comparative regression tables

Table 1: OLS regression of Morris Index, by country. (Household level) 61

Table 2: OLS regression on food insecurity index, by country (Household level) 63

Table 3: Comparative analysis on access to health services (Household level) 65

Table 4: Comparative analysis on access to education services (Household level) 67

Table 5: Comparative analysis on access to water services (Household level) 70

Table 6: Comparative analysis on access to social protection (Household level) 72

Table 7: Comparative analysis on access to livelihood services (Household level) 73

Table 8: Comparative analysis on experience of health services. (Satisfied) 75

Table 9: Comparative analysis on experience of health services. (Dissatisfied) 78

Table 10: Comparative analysis on experience of education service: Satisfied. 81

Table 11: Comparative analysis on experience of water services: Water is clean and safe. 83

Table 12: Comparative analysis on experience of social protection. 86

Table 13: Comparative analysis on experience of livelihood services. 88

Table 14: Comparative analysis: Local government cares about opinions 89

Table 15: Comparative analysis: Local government decisions reflect respondent’s priorities (Never or almost never). 92

Table 16: Comparative analysis: Local government decisions reflect respondent’s priorities (Very much/ to a large extent or Always). 95

Table 17: Comparative analysis: Central government cares about opinions. 98

Table 18: Comparative analysis: Central government decisions reflect respondent’s priorities (Never or almost never). 101

Table 19: Comparative analysis: Central government decisions reflect respondent’s priorities (Very much/ to a large extent or Always). 104

61

Table 1: OLS regression of Morris Index, by country. (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female-headed

household -9.36** -4.96*** n/a -0.47*** -0.56*

Average age of

household 0.12 0.13*** 0.13 0.00 0.05***

Main activity of

household

-7.59** (main income

source: agriculture)

2.38*** (main income

source: own cultivation)

9.45*** (main

income source:

agriculture)

0.23 (main activity: fishing)

0.26 (main activity: agriculture)

-0.46*** (main activity: trading)

Main income source:

-0.04 (casual agric.)

0.11 (casual non agric.)

0.60 (exploitation of bush

products)

1.29 (other assistance)

0.29 (other economic activity)

2.64*** (own business)

-0.25 (own business, market)

-0.60 (own fishing)

2.10** (own livestock)

-0.33 (paid housework)

-0.12 (private/ NGO)

2.27** (government)

(ref: none)

Education level of the

adult household

members

5.97*** (mean years of

education)

4.21*** (median

education of adults)

2.05*** (Average

education of adults)

0.88*** (share of adults

completing primary)

Respondent education:

-0.22 (some primary)

0.71* (finished primary)

0.80* (0-level)

2.26*** (A-level)

1.68** (tertiary)

(ref: none)

Household has migrant n/a (Only 5 households

had a current migrant) -3.72** 9.54** * 0.36*** 1.66

Household receives

remittances 1.28 1.75* 4.19 0.03 0.97**

Household has been

displaced

-10.43**(displaced due to

conflict)

-0.92 (lived in village all

life)

1.64 (displaced due

to conflict) -0.11 (ever been displaced) -0.21 (ever been displaced)

Religion 4.76 (Religious minority) n/a n/a n/a n/a

Ethnicity -7.64* (Ethnic minority) -2.28** (minority) n/a -0.02 (minority)

0.82* (Langi)

-3.76*** (Iteso)

-1.54** (Kumam)

-1.80** (Karamojong)

-0.97 (Mixed)

-2.61*** (Other)

(ref: Acholi)

Dependency ratio -1.01 -1.47** -1.05 -0.16 0.00***

62

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Context

Urban n/a 2.63*** n/a 0.79*** 0.79*

Household has

experienced conflict in

past three years

-11.92*** -0.69 n/a n/a -0.93

Perception of safety -4.68 (safe in village)

8.29** (safe at home)

1.89 (safe in

neighbourhood)

-6.41*** (safe when

moving around)

0.91*** (safe in

neighbourhood) -0.44*** (safe in neighbourhood)

Household has access to

credit -0.88 4.83***

-10.55***

(household has debt) 0.35** 0.78**

Location n/a

7.41*** (Rolpa)

8.94*** (Bardiya)

(ref: Ilam)

4.49*** (Swat)

(ref:Lower Dir) n/a

-0.30 (Sub-region: Langi) (ref:

Acholi)

Shocks

Shocks experienced 1.16 0.07 6.73*** -0.12*** 0.10***

Crimes experienced 1.88 -0.03 3.31*** 0.19* 0.01 (crimes)

-0.01 (serious crimes)

Access to services

and social protection

Access to health 0.01 0.02* 0.02 0.00*** 0.00

Access to water services -0.19*** 0.03 n/a 0.00 -0.01**

Access to social

protection -3.62 1.93** 0.63 -0.13*** 0.40

Access to livelihood

services 9.24** 3.21*** 8.42*** 0.61*** 1.04**

Quality of health -0.94 (Satisfied) -0.67 (satisfied) 6.06***(satisfied) 0.23* (satisfied)

0.41 (fairly satisfied)

0.03 (dissatisfied)

(ref: satisfied)

Quality of water clean -0.13 -0.04 -2.52 -0.25 0.36

Information about the

regression model

R2 0.11 0.15 0.26 0.28 0.21

Number of observations 657 2812 2085 1375 1519

63

Table 2: OLS regression on food insecurity index, by country (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female-headed

household 1.01 -0.81** n/a 1.78*** -0.34

Average age of

household -0.10*** -0.05*** -0.02 -0.07*** 0.04

Main activity of

household

0.59 (main income source:

agriculture)

0.13 (main income source:

own cultivation)

-1.04*** (main

income: farming)

-0.52 (main income:

overseas labour)

0.63 (main activity: fishing)

-0.57 (main activity:

agriculture)

-0.52 (main activity:

trading)

Main income source:

7.01** (casual agric.)

4.53** (casual non agric.)

3.04 (exploitation of bush

products)

4.56 (other assistance)

1.17 (other economic activity)

-2.93 (own business)

-2.99 (own business, market)

-9.53** (own fishing)

-2.99** (own livestock)

-1.94 (paid housework)

-3.34** (private/ NGO)

-4.88** (government)

(ref: none)

Education level of the

adult household

members

-0.34 (mean years of education) -0.64*** (median education

of adults)

-0.22*** (Average

education of adults)

-1.92*** (share of adults

completing primary)

Respondent educ:

-3.80*** (some primary)

-3.42**(primary)

-3.01** (O level)

-2.82 (A level)

-5.45*** (Tertiary)

(ref: none)

Household has migrant n/a 0.57 -1.16** 0.58 2.63

Household receives

remittances -0.95 -0.36 0.54 0.53 0.96

Household has been

displaced

2.26** (displaced due to

conflict)

0.10 (lived in village entire

life)

0.06 (displaced due to

conflict) 0.84 (ever displaced) 0.72 (ever displaced)

Religion -2.68*** (minority) n/a n/a n/a n/a

Ethnicity 0.35 (minority) 1.39*** (minority) n/a 0.57 (minority)

-2.21 (Langi)

-18.58*** (Iteso)

-6.06** (Kumam)

-14.00*** (Karamojong)

-1.45 (Mixed)

-10.12*** (Other)

(ref: Acholi)

Dependency ratio 0.24 0.07 0.3** 0.25 n/a

Morris Index -0.02** -0.03*** -0.01*** -0.66*** -0.44***

64

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Context

Urban n/a 1.21*** n/a 0.18 -0.46

Household has

experienced conflict in

past three years

1.96*** 0.05 n/a n/a 1.59

Perception of safety 1.33** (safe in village)

-0.59 (safe while travelling)

-0.98*** (safe in

neighbourhood)

0.24 (safe when

moving around)

0.59 (safe in

neighbourhood)

1.97*** (safe in

neighbourhood)

Household has access to

credit 1.72*** -2.62*** 1.39*** (owes money) -1.26** -2.05**

Location n/a

0.38 (Rolpa)

0.84*** (Bardiya)

(Ref: Ilam)

1.76*** (Swat)

(ref: Lower Dir) n/a

2.89* (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced 0.27* 0.04 0.29** 0.71*** 0.10

Crimes experienced 1.59*** 0.10*** -0.46*** 1.09** -0.08* (crimes)

0.26*** (serious crimes)

Access to services

and social

protection

Access to health -0.01 0.01*** -0.02*** 0.00 0.00

Access to water services 0.02 0.00 n/a 0.03** -0.01

Access to social

protection 0.62 0.52*** 0.65*** 1.04** -0.39

Access to livelihood

services 0.50 -0.32 -0.45* -0.16 -2.33***

Quality of health -2.01***(satisfied) 0.26** (satisfied) 0.72*** (satisfied) -0.89** (satisfied)

0.32 (fairly satisfied)

0.03 (dissatisfied)

(ref: satisfied)

Quality of water clean 0.32 -1.31*** -0.27 -1.27** -1.72**

Information about the

regression model

R2 0.28 0.19 0.15 0.13 0.14

Number of observations 650 2812 2085 1375 1519

65

Table 3: Comparative analysis on access to health services (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female-headed household -3.54 -0.96 1.30 (female respondent) 3.26 -11.42*

Average age of household -0.08 0.02 -0.08 -0.34*** 0.19

Main activity of household 6.36 (main income source:

agriculture)

9.13*** (main income

source: own cultivation)

0.84 (HH head in farming)

2.04 (main income:

agriculture)

3.16 (main income:

overseas labour)

7.26 (main activity: fishing)

-0.64 (main activity: agriculture)

1.13 (main activity: trading)

Main income source:

-7.14 (casual labour)

-21.73 (casual non-

agriculture)

-12.73 (exploitation of

bush products)

22.58 (other assistance)

12.37 (other economic

activity)

-10.26 (shop-building)

-18.64 (trading)

210.55*** (fishing)

-10.40 (livestock)

28.6** (paid housework)

10.51 (private/ NGO)

27.04 (government)

(ref: none)

Education level of the adult

household members

-3.25* (mean years of

education)

-4.12*** (median

education of adults)

-0.15 (Average education of

adults)

-8.69** (Share of adults

completing primary)

Respondent educ:

19.01** (some primary)

12.17 (primary)

1.42 (O level)

-43.24*** (A level)

-21.10* (Tertiary)

(ref: none)

Household has migrant n/a 2.46 8.03* (internal)

-2.28 (international) -8.33*** -18.14

Household receives

remittances 5.09 5.41*** 0.29 -0.38 -5.67**

Household has been

displaced

0.42 (displaced due to

conflict)

-3.72** (lived in village

whole life)

7.55*** (displaced due to

conflict) 10.06** (ever displaced) 4.08 (ever displaced)

Religion 10.47 (minority) n/a n/a n/a n/a

Ethnicity -5.82 (minority) -5.02** (minority) n/a 20.94***(minority)

-4.06 (Langi)

-54.71*** (Iteso)

23.35 (Kumam)

3.63 (Karamojong)

-32.37** (Mixed)

-14.11 (Other)

(ref: Acholi)

Dependency ratio -1.39 0.84 0.39 1.13 0.07**

66

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Morris Index -0.03 0.11** -0.73*** -3.66*** -0.96

Food Security Index -0.24 0.59*** 0.02 0.17 0.13

Context

Urban -2.67 -17.22*** -31.88***

Household has

experienced conflict in past

three years

-23.74*** -2.96** 2.26 n/a 8.59

Perception of safety -0.64 (safe in village)

5.60 (safe while travelling)

-3.32 (safe in

neighbourhood)

1.16 (safe when moving

around) -8.42 (safe in neighbourhood)

-13.49*** (safe in

neighbourhood)

Private form of

transportation -9.10*

6.14*** (man-powered

vehicle)

-4.39 (petrol powered

vehicle)

-4.00 (motorbike)

1.43 (car) -23.14*** (satisfied with access

to transport)

-1.78 (bicycle)

-35.74 (cart)

-14.34 (car)

Access to public

transportation -9.27 n/a n/a n/a

Location n/a

17.33*** (Rolpa)

-26.26*** (Bardiya)

(ref: Ilam)

-5.57*** (Swat)

(ref: Lower Dir) n/a

30.82** (Sub-region:

Langi) (reference: Acholi)

Shocks

Shocks experienced 1.24 0.78*** -0.42 2.13** -0.81

Crimes experienced -2.71 -0.31 -3.61*** 0.49 -0.10 (crimes)

2.24*** (serious crimes)

Perception of

health services

Number of staff

1.48 (neutral)

0.64 (satisfied)

(ref: dissatisfied)

1.40 (satisfied)

-0.23 (dissatisfied)

(ref: neutral)

4.41*** (satisfied) n/a

16.36* (fairly satisfied)

-11.08 (dissatisfied)

(ref: satisfied)

Availability of medicine

2.27 (neutral)

-3.13 (satisfied)

(ref: dissatisfied)

2.25 (satisfied)

-5.61 (dissatisfied)

(ref: neutral)

0.14 (satisfied) n/a n/a

Waiting times

-2.72 (neutral)

-11.30* (satisfied)

(ref: dissatisfied)

-7.93*** (satisfied)

-0.02 (dissatisfied)

(ref: neutral)

2.95*** (satisfied) n/a

-2.45 (fairly satisfied)

36.92*** (dissatisfied)

(ref: satisfied)

Quality of equipment n/a n/a n/a 9.44***

4.22 (fairly satisfied)

12.85 (dissatisfied)

(ref: satisfied)

Satisfied with overall

quality n/a n/a n/a 6.15** n/a

Formal Health Costs -14.00 11.37*** -1.77 -10.75 18.6*

Informal Health Costs 13.13** 3.44 -0.21 13.31** -2.42

67

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Provider of health facilities -2.92 (government) 4.64** (government) 2.73 (Not run by

government) -21.13*** (Run by government)

-14.94 (privately run)

22.24 (religious

organisation)

-56.43** (NGO)

11.14 (other)

(ref: government)

Community Meeting Held

on Health -1.45 -2.11 11.11*** n/a -0.31

Attended Community

Meeting on Health -0.74 7.03** 16.99* -1.39 n/a

Information about

the regression model

R2 0.15 0.24 0.09 0.55 0.15

Number of observations 527 2973 2114 1352 1438

Table 4: Comparative analysis on access to education services (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female-headed household 8.34 3.72 2.36** (female

respondent) 4.04 -2.18

Average age of household 0.06 -0.18 -0.11 -0.44 -0.23**

Main activity of household 16.48** (main income

source: agriculture)

2.66 (main income source:

own cultivation)

7.17* (farming is main

activity)

0.69 (main income:

farming)

-0.32 (main income:

overseas labour)

0.99 (main activity: fishing)

1.60 (main activity:

agriculture)

-1.18 (main activity: trading)

Main income source:

-0.46 (casual labour: agr)

13.29* (casual labour: non-

agr)

6.93 (exploitation of bush

products)

4.44 (other assistance)

-7.15 (other economic activity)

1.55 (own business: shop

building)

-2.16 (business: home,

market)

-11.34** (livestock)

-11.73 (paid housework)

4.64 (private/NGO)

-7.74 (government)

(ref: none)

Education level of the adult

household members

0.73 (mean years of

education) -0.88 (respondent education)

0.00 (Average education

of adults)

5.19 (share completing

primary)

Respondent education

-2.69 (some primary)

-7.24 (primary)

-17.18*** (O level)

-15.43* (A level)

-21.01*** (Tertiary)

(ref: none)

68

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household has migrant n/a 5.20

-3.88* (internal migrant)

-2.81 (international

migrant)

-2.89 2.52

Household receives remittances 5.66 3.41 1.33 8.71 -2.17

Household has been displaced 9.70 (displaced due to

conflict) 1.32 (lived in village all life)

-0.98 (displaced due to

conflict) 5.59 (ever displaced) 6.77 (ever displaced)

Religion 8.97 (minority) n/a n/a n/a n/a

Ethnicity -10.17 (minority) -3.51(minority) n/a 13.08*** (minority) n/a

Dependency ratio -1.88 0.28 -3.24* -6.01** 0.02

Morris Index -0.04 0.08 0.01 -3.01* -0.54*

Food security index -0.54 0.08 0.12 0.29 -0.23*

Context

Urban n/aa 6.00** n/a n/a -18.35***

Household has experienced conflict

in past three years -24.45*** -1.15 -1.95 n/a -16.09

Perception of safety 5.28 (safe in village)

3.78 (safe while t ravelling) 3.45 (safe in neighbourhood)

0.78 (safe when moving

around) -7.97 (safe in neighbourhood) 1.85 (safe in neighbourhood)

Private form of transportation 4.30 -1.36 (man-powered vehicle)

-4.98 (petrol-powered vehicle)

0.27 (motorbike)

3.56** (car)

-45.41*** (satisfied with

transport: boys)

30.26*** (satisfied with

transport: girls)

1.33 (bicycle)

-23.76** (cart)

-6.40 (car)

Access to public transportation 13.08 n/a n/a n/a n/a

Location

4.52 (Rolpa)

-10.13*** (Bardiya)

(ref: Ilam)

-1.24 (Swat, reference:

Lower Dir) n/a

4.31 (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced -0.14 0.27 0.22 1.68 0.08

Crimes experienced -3.80 -0.18 -0.70 6.67 -0.15 (crimes)

0.23 (serious crimes)

Perception of education Number of teachers

1.06 (dissatisfied)

9.56 (neutral)

(ref: satisfied)

1.52 (satisfied: boys)

-31.15* (dissatisfied: boys)

-0.82 (satisfied: girls)

31.35** (dissatisfied: girls)

(ref: neutral)

6.38 (neutral: boys)

4.33 (satisfied: boys)

0.44 (neutral: girls)

3.83 (satisfied: girls)

(ref: dissatisfied)

4.28 (satisfied: boys)

4.44 (satisfied: girls)

-0.62 (fairly satisfied)

4.05 (dissatisfied)

(ref: satisfied)

Quality of teachers -10.67 (dissatisfied) 6.48 (satisfied: boys) n/a n/a -1.22 (fairly satisfied)

69

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

-13.22 (neutral)

(ref: satisfied)

3.81 (dissatisfied: boys)

-8.92 (satisfied: girls)

3.84 (dissatisfied: girls)

(ref: neutral)

5.61 (dissatisfied)

(ref: satisfied)

Attendance of teachers

-0.17 (dissatisfied)

2.58 (neutral)

(ref: satisfied)

0.22 (satisfied: boys)

10.36 (dissatisfied: boys)

1.03 (satisfied: girls)

-9.67 (dissatisfied: girls)

(ref: neutral)

0.35 (neutral: boys)

-2.58 (satisfied: boys)

2.22 (neutral: girls)

0.82 (satisfied: girls)

(ref: dissatisfied)

n/a

1.99 (fairly satisfied)

5.14 (dissatisfied)

(ref: satisfied)

Size of class

0.30 (dissatisfied)

5.64 (neutral)

(ref: satisfied)

-1.58 (satisfied: boys)

-15.94 (dissatisfied: boys)

3.57 (satisfied: girls)

-3.45 (dissatisfied: girls)

(ref: neutral)

5.26** (neutral: boys)

3.24 (satisfied: boys)

-8.83*** (neutral: girls)

-4.99* (satisfied: girls)

(ref: dissatisfied)

n/a

-1.55 (fairly satisfied)

9.92* (dissatisfied)

(ref: satisfied)

School infrastructure

-1.74 (dissatisfied)

-16.91* (neutral)

(ref: satisfied)

-4.81 (satisfied: boys)

4.25 (dissatisfied: boys)

9.13 (satisfied: girls)

-5.14 (dissatisfied: girls)

(ref: neutral)

-0.29 (neutral: boys)

3.31 (satisfied: boys)

-0.27 (neutral: girls)

-2.94 (satisfied: girls)

(ref: dissatisfied)

n/a

-1.60 (fairly satisfied)

-9.21* (dissatisfied)

(ref: satisfied)

Quality of equipment

-5.83 (dissatisfied)

6.92 (neutral)

(ref: satisfied)

5.60 (satisfied: boys)

-5.98 (dissatisfied: boys)

-10.47 (satisfied: girls)

8.55 (dissatisfied: girls)

(ref: neutral)

n/a -33.79** (satisfied: boys)

34.51** (satisfied: girls)

2.27 (fairly satisfied)

6.50 (dissatisfied)

(ref: satisfied)

Official Fees for School -1.25 -9.18 (boys)

5.63 (girls)

-2.31 (boys)

3.09* (girls)

38.79** (boys)

-29.30 (girls) n/a

Informal fees for school n/a -0.35 (boys)

6.16 (girls)

-10.20 (boys)

3.73 (girls)

26.79* (boys)

-20.54 (girls) n/a

Provider 1.78 (government) -3.69 (government: boys)

0.00 (government: girls) n/a

-29.26 (government: boys)

3.54 (government: girls)

3.90 (private)

8.06 (religious)

65.00** (NGO)

-19.15** (other)

(ref: government)

Community Meeting Held on

Education -18.43* -6.33 0.47 n/a -0.70

Attended Community Meeting on

Education 5.29 5.06 n/a 0.86 n/a

Information about the

regression model

R2 0.17 0.23 0.14 0.32 0.19

Number of observations 357 487 535 316 776

70

Table 5: Comparative analysis on access to water services (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household

factors

Female-headed household 0.07 0.24 0.30 (female respondent) -0.23 6.27

Average age of household 0.05 -0.02 -0.00 0.00 0.03

Main activity of household -2.74 (main income source:

agriculture)

0.25 (main income source:

own cultivation)

-1.01 (HH head in farming)

-0.03 (main income: farming)

-0.61 (main income: overseas

labour)

0.09 (main activity: fishing)

0.38* (main activity:

agriculture)

0.50*** (main activity:

trading)

Main income source:

-0.35 (casual labour: agr)

15.1 (casual labour: non-agr)

6.37 (exploitation of bush

products)

97.21** (other assistance)

-2.61 (other economic activity)

-9.83 (own business: shop)

-6.71 (own business: home,

market)

12.04 (fishing)

-2.76 (livestock)

-19.30*** (paid housework)

-3.49 (private/ NGO)

4.46 (government)

(ref: none)

Education level of the adult

household members

2.13** (mean years of

education)

-0.17 (median education of

adults)

-0.17*** (Average education

of adults)

-0.32** (share of adults

completing primary)

Respondent educ:

1.78 (some primary)

4.82 (primary)

0.95 (O level)

-5.64 (A level)

-0.30 (Tertiary)

(ref: none)

Household has migrant n/a 0.30 0.58 (internal)

-2.93*** (international) -0.34*** -12.36***

Household receives

remittances 2.02 0.43 3.13*** -0.09 -3.31**

Household has been

displaced

-6.19** (displaced due to

conflict) -0.54 (lived in village all life)

-0.58 (displaced due to

conflict) 0.39** (ever displaced) -6.74** (ever displaced)

Religion 0.16 (minority) n/a n/a n/a n/a

Ethnicity -4.94** (minority) 0.48 (minority) n/a -1.55*** (minority)

8.23* (Langi)

45.11*** (Iteso)

0.96 (Kumam)

75.89 (Karamojong)

-5.02 (Mixed)

-2.83 (Other)

(ref: Acholi)

Dependency ratio -0.53 0.44 -0.02 0.20** 0.02*

71

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Morris Index -0.04** 0.00 0.01 -0.01 -0.65*

Food Security Index 0.22 0.05 -0.12*** -0.02** -0.13

Context

Urban n/a -0.24 n/a -3.33*** -7.34

Household has experienced

conflict in past three years -6.24*** 0.53 0.73 n/a -11.94*

Perception of safety 1.65 (safe in village)

-1.43 (safe while travelling) 0.04 (safe in neighbourhood)

-0.84* (safe when moving

around) 0.04 (safe in neighbourhood) 3.02 (safe in neighbourhood)

Private form of transportation -0.36 0.59 (man-powered)

-0.06 (petrol-powered)

-0.77 (motorbike)

0.39 (car) 0.21** (satisfied with access

to transport)

1.34 (bicycle)

2.09 (cart)

-8.94** (car) Access to public

transportation -2.50 n/a n/a

Location n/a

12.13 (Rolpa)

-2.57 (Bardiya)

(ref: Ilam)

-2.80*** (Swat) n/a 4.67 (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced 0.31 0.05 0.33 -0.15*** 0.58*

Crimes experienced 1.12 0.07 0.74** -0.36** -0.37*** (crimes)

0.34 (serious crimes)

Perception

of health

services

Have to Queue for Water 4.93** 2.66 0.01* 0.99*** 6.64**

Official Fees for Water -13.55*** -0.54 -0.00 0.31** 4.33

Who runs water

1.68 (committee)

0.14 (NGO)

4.15 (nobody)

(ref: government)

-6.21 (committee)

-6.42 (NGO)

-8.98 (nobody)

(ref: government)

-0.28 (not run by government) 0.78*** (government)

3.65 (government)

-11.32*** (private)

-0.67 (community)

2.90 (NGO)

-8.25 (other)

(ref: nobody)

Community Meeting Held on

Water -4.17 2.67 -0.80 n/a -4.86*

Attended community Meeting

on Water 4.71* -1.72 -0.66 0.60*** n/a

Information R2 0.12 0.33 0.06 0.55 0.11

72

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

about the

regression

model

Number of observations 745 2982 2114 1374 1478

Table 6: Comparative analysis on access to social protection (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female-headed household 0.09 0.97*** -0.10 (respondent

female) 0.47 -0.19

Average age of household -0.01 0.01*** -0.04*** 0.02* 0.00

Main activity of household -0.34* (main income

source: agriculture)

0.23** (main income

source: own

cultivation)

1.08** (HH head in

farming)

-0.39** (main

income: farming)

-0.27 (main income:

overseas labour)

1.72*** (main activity:

fishing)

0.99** (main activity:

agriculture)

0.82** (main activity:

trading)

Main income source:

0.10 (exploitation of bush products)

0.83 (other economic activity)

-0.99 (business: home, market)

0.73 (own livestock)

0.13 (private/ NGO)

-0.03 (work for government)

(ref: none)

Education level of the adult

household members

-0.11 (mean years of

education)

-0.22*** (median

education of adults)

-0.17*** (average

education of adults)

0.03 (share of adults

completing primary)

Respondent educ:

0.80** (some primary)

0.01 (primary)

0.61 (O level)

0.16 (A level)

-0.09 (tertiary)

(ref: none)

Household has migrant n/a 0.50*** 0.07 (internal)

-0.45 (international) 0.46** 1.09**

Household receives remittances 0.35 -0.38*** 0.57* -0.22 0.24

Household has been displaced 0.46** (displaced

due to conflict)

0.03 (lived in village

all life)

-0.12 (displaced due

to conflict) -0.25 (ever displaced) 0.43 (ever displaced)

Number of children -0.01 0.46*** 0.13 (dependency

ratio) 0.17** 0.00

Number of elderly -0.34 0.64*** n/a 0.02 0.61**

Religion -0.01 (minority) n/a n/a n/a n/a

Ethnicity 0.66*** (minority) 0.73*** (minority) n/a 0.34 -0.45 (Langi)

Morris Index -0.00 0.00 0.02* -0.07 0.01

73

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Food Security Index 0.01 0.02* -0.00 0.03** -0.01

Context

Urban n/a 0.10 n/a 2.68*** 0.25

Household has experienced

conflict in past three years 0.97*** 0.12 0.14 n/a n/a

Perception of safety

0.13 (safe in village)

0.32 (safe while

travelling)

0.07 (safe in

neighbourhood) n/a

1.09 (safe in

neighbourhood) 0.11 (safe in neighbourhood)

Location n/a

0.25**(Rolpa)

0.16 (Bardiya)

(ref: Ilam)

1.23*** n/a 0.25 (Sub-region: Langi) (reference: Acholi)

Shocks

Shocks experienced -0.03 -0.02 0.17** 0.09 -0.03

Crimes experienced 0.14 0.00 -0.08 0.28 -0.01 (crimes)

0.00 (serious crimes)

Community Meeting Held

(Social Protection) n/a 0.16 -0.27 n/a 1.23**

Attended Meeting on Social

Protection n/a 0.23 -0.22 0.00 n/a

Information about

the regression model

R2 0.13 0.15 n/a 0.07 n/a

Number of observations 933 3051 2114 1375 1493

Table 7: Comparative analysis on access to livelihood services (Household level)

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household

factors

Female-headed household 0.03 0.04 0.01 (female respondent) 0.10 -0.10

Average age of household -0.01 -0.02*** 0.01 0.00 0.00

Main activity of household -0.57** (main income source:

agriculture)

0.17 (main income source:

own cultivation)

1.16* (HH head in farming)

0.38* (main income: farming)

-0.53*(main income:

overseas)

1.50*** (main activity:

fishing)

2.52*** (main activity:

agriculture)

0.41 (main activity: trading)

Main income source:

-0.77 (casual labour: agr)

-1.12 (casual labour: non-agr)

-0.93 (exploitation of bush

products)

1.3 (other assistance)

-0.30 (other economic

activity)

-0.26 (own business: shop

building)

-0.09 (business: home,

market)

1.22*** (own livestock)

74

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

0.71 (private/NGO)

-0.55 (government)

(ref: none)

Education level of the adult

household members

-0.11 (mean years of

education)

0.07 (median education of

adults)

-0.01 (average education of

adults)

0.20 (Share of adults

completing primary)

Respondent educ:

0.00 (some primary)

0.24 (primary)

0.25 (O level)

-0.92 (A level)

-0.01 (Tertiary)

(ref: none)

Household has migrant n/a -0.04 -0.16 (internal)

0.23 (international) 0.45 0.27

Household receives

remittances 0.79** 0.25* -0.06 0.16 0.20

Household has been

displaced

0.51* (displaced due to

conflict) -0.06 (lived in village all life)

-0.31 (displaced due to

conflict) 0.29 (ever displaced) 0.42 (ever displaced)

Dependency ratio 0.03 -0.14 -0.09 0.04 0.00

Household owns land 0.13 0.33 n/a -0.05 0.03**

Religion -0.70* (minority) n/a n/a n/a n/a

Ethnicity -0.65*** (minority) 0.03 (ethnicity) n/a 0.20 (minority)

-0.36 (Langi)

-0.47 (Kumam)

0.60 (Mixed)

-0.96 (Other)

(ref: Acholi))

Morris Index 0.00* 0.01*** -0.02 0.33*** 0.03

Food Security Index 0.00 0.00 0.01*** -0.01 -0.02***

Context

Urban n/a -0.08 n/a -1.04*** -0.26

Household has experienced

conflict in past three years 1.83*** 0.03 0.27 n/a -0.94

Perception of safety -0.31 (safe in village)

-0.02 (safe while travelling) -0.09 (safe in neighbourhood)

0.00 (safe when moving

around) 0.04 (safe in neighbourhood) 0.07 (safe in neighbourhood)

Location n/a

0.70***(Rolpa)

-0.41***(Bardiya)

(ref: Ilam)

1.85*** (Swat)

(ref: Lower Dir) n/a

0.87 (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced 0.04 -0.02 0.01 0.33*** 0.04**

Crimes experienced 0.13 0.03* 0.22*** 0.24 0.00 (crimes)

0.01 (serious crimes)

75

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Community Meeting Held

(Livelihood Services) n/a 0.25 -0.69 n/a 1.26***

Attended Meeting on

Livelihood n/a 1.41*** 1.02 0.65*** n/a

Information

about the

regression

model

R2 0.14 0.09 0.24 0.24 n/a

Number of observations 920 3005 2087 1377 1430

Table 8: Comparative analysis on experience of health services. (Satisfied)

Satisfied Explanatory variable DRC (satisfied) Nepal Pakistan (satisfied) Sri Lanka

(satisfied) Uganda (satisfied)

Household factors

Female respondent 0.11 (Satisfied is the base

category) -0.39** -0.19 (male) 0.43* (female head)

Age of respondent -0.00 n/a -0.01 0.00 0.01

Main activity of household -0.23 (main income source:

agriculture)

-0.40 (main income:

farming)

-0.25 (main income:

overseas labour)

0.18 (respondent is

employed)

Main income source:

0.99 (casual labour: agr)

1.58** (casual labour: non-agr)

-0.59 (exploitation of bush products)

2.49** (other assistance)

0.79 (other economic activity)

-0.10 (own business: shop building)

1.06*** (business: home, market)

23.84*** (fishing)

0.36 (livestock)

22.34*** (paid housework)

0.51 (private/ NGO)

0.80 (work for government)

(ref: none)

Education level of respondent -0.13 (categorical) -0.02 (average education of

adults)

0.22 (respondent

completed primary)

Respondent educ:

-0.03 (some primary)

0.02 (primary)

-0.45 (O level)

-0.04 (A level)

0.05 (tertiary)

(ref: none)

Household has migrant n/a n/a -0.26 0.64

Household receives

remittances -0.16 0.35 0.50* -0.13

76

Satisfied Explanatory variable DRC (satisfied) Nepal Pakistan (satisfied) Sri Lanka

(satisfied) Uganda (satisfied)

Household has been

displaced 0.41 (displaced due to conflict)

-0.37 (displaced due to

conflict) -0.65* (ever displaced) -0.59** (eve displaced)

Religion 0.08 n/a n/a n/a

Ethnicity -0.25 (minority) n/a -0.20 (ethnicity)

-0.42 (Langi)

-18.64*** (Iteso)

-1.20 (Kumam)

-22.04*** (Karamojong)

1.81* (Mixed)

-0.92 (Other)

(ref: Acholi)

Dependency ratio 0.08 -0.16 0.24 0.00

Morris Index 0.00 0.01 0.01 -0.04

Food Security Index -0.04* -0.01 -0.01 0.00

Context

Urban n/a n/a 0.83*** -0.17

Household has experienced

conflict in past three years -0.14 0.09 n/a -0.65

Perception of safety -0.43 (safe in village)

0.72* (safe while travelling)

0.33 (safe when moving

around)

-0.12 (safe in

neighbourhood) -0.19 (safe in neighbourhood)

Location n/a 1.95*** n/a 0.06 (Sub-region: Langi) (reference:

Acholi)

Shocks

Shocks experienced -0.02 -0.12 0.13* -0.01

Crimes experienced 0.18 0.17 0.02 (crimes)

0.01 (serious crimes)

Perception of

health services

Distance to health centre -0.00

n/a n/a -0.00**

Number of staff

2.03*** (neutral)

1.36*** (satisfied)

(ref: dissatisfied)

0.75 (satisfied) n/a

-2.37*** (fairly satisfied)

-1.51*** (dissatisfied)

(ref: satisfied)

Availability of medicine

0.85***(neutral)

1.64*** (satisfied)

(ref: dissatisfied)

2.38*** (satisfied) 2.11*** (satisfied)

Waiting times

1.39***(neutral)

1.43***(satisfied)

(ref: dissatisfied)

1.06*** (satisfied) 2.50*** (satisfied)

-1.17*** (fairly satisfied)

-1.73*** (dissatisfied)

(ref: satisfied)

Quality of equipment n/a n/a n/a

-1.85*** (fairly satisfied)

-1.76*** (dissatisfied)

(ref: satisfied)

77

Satisfied Explanatory variable DRC (satisfied) Nepal Pakistan (satisfied) Sri Lanka

(satisfied) Uganda (satisfied)

Formal Health Costs 2.60** -0.17 -0.01 0.53

Informal Health Costs -0.75* n/a -0.11 -0.42

Provider of Health Facilities -0.51 (government) 0.45 (service not run by

government) 0.03 (government)

1.43*** (private)

0.76 (religious)

1.38 (NGO)

-19.73*** (other)

(ref: government)

Community Meeting Held on

Health 0.94** n/a n/a -0.1

Attended Community Meeting

on Health -1.59*** n/a 0.01 n/a

Information about the

regression model

R2 0.45 0.65 0.39 n/a

Number of observations 512 2062 1352 1407

78

Table 9: Comparative analysis on experience of health services. (Dissatisfied)

Explanatory

variable Nepal (neutral) Nepal (dissatisfied)

Pakistan

(Indifferent)

Pakistan (very

dissatisfied) Sri Lanka

Uganda

(dissatisfied)

Household

factors

Female respondent -0.07 -0.28 0.64*** 0.73* (Dissatisfied is the

base category) 0.17 (female head)

Age of respondent 0.00 -0.01 0.02 0.00 0.01

Main activity of

household

0.21 (main income

source: own cultivation)

0.34 (main income source:

own cultivation)

-1.03*** (main income:

farming)

-0.78 (main income:

overseas labour)

-0.55 (main income:

farming)

-0.80 (main income:

overseas labour)

Main income source:

0.56 (casual labour: agr)

0.14 (casual labour: non-agr)

-0.13 (exploitation of bush

products)

-20.00*** (other

assistance)

2.43*** (other economic

activity)

-0.49 (own business: shop

building)

-0.12 (business: home,

market)

0.61 (fishing)

0.48 (livestock)

0.43 (paid housework)

-0.57 (private/ NGO)

-1.25 (work for government)

(ref: none)

Education level of

respondent 0.00 (categorical) -0.04 (categorical)

-0.01 (average education

of adults) -0.02

Respondent educ:

-0.02 (some primary)

0.00 (primary)

0.34 (O level)

-0.82 (A level)

-0.01 (tertiary)

(ref: none)

Household has

migrant -0.41 0.56 1.08

Household receives

remittances -0.10 -0.06 0.21 0.3 0.13

Household has been

displaced

0.21 (lived in village all

life)

-0.05 (lived in village all

life)

-0.58 (displaced due to

conflict)

0.21 (displaced due to

conflict) 0.27 (ever displaced)

Ethnicity 0.00 (minority) -0.17 n/a n/a

0.01 (Langi)

-22.95*** (Iteso)

0.11 (Kumam)

-19.61*** (Karamojong)

-0.03 (Mixed)

-21.02 (Other)

79

Explanatory

variable Nepal (neutral) Nepal (dissatisfied)

Pakistan

(Indifferent)

Pakistan (very

dissatisfied) Sri Lanka

Uganda

(dissatisfied)

(ref: Acholi)

Dependency ratio 0.04 -0.37* -0.08 -0.25 0.00

Morris Index 0.01 0.00 0.01 -0.04 -0.04

Food Security Index 0.04*** 0.04 0.02 0.02 0.00

Context

Urban -0.45** -0.54* n/a n/a -0.14

Household has

experienced conflict

in past three years

0.32** 0.20 -0.03 0.41 0.32

Perception of safety -0.22 (safe in

neighbourhood)

-0.74* (safe in

neighbourhood)

0.17 (safe when moving

around)

-0.45 (safe when moving

around)

-0.14 (safe in

neighbourhood)

Location

-0.51** (Rolpa)

-0.22 (Bardiya)

(ref: Ilam)

-1.24** (Rolpa)

-1.04*** (Bardiya)

(ref: Ilam)

0.92*** (Swat)

(ref: Lower Dir)

-2.22*** (Swat)

(ref: Lower Dir)

0.26 (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced 0.05*** 0.08*** 0.26** -0.17 0.02

Crimes experienced -0.01 0.01 -0.37 0.07 0.00 (crimes)

-0.02 (serious crimes)

Perception of

health services

Distance to health

centre 0.00 0.00 n/a n/a 0.00

Number of staff

-2.58*** (satisfied)

-0.11 (dissatisfied)

(ref: neutral)

-2.28*** (satisfied)

2.31*** (dissatisfied)

(ref: neutral)

0.65*** (satisfied) -2.92*** (satisfied)

0.19 (fairly satisfied)

1.49*** (dissatisfied)

(ref: satisfied)

Availability of

medicine

-2.22*** (satisfied)

-0.06 (dissatisfied)

(ref: neutral)

-2.03*** (satisfied)

1.91*** (dissatisfied)

(ref: neutral)

0.68*** (satisfied) -1.35* (satisfied) n/a

Waiting times

-1.41*** (satisfied)

0.25 (dissatisfied)

(ref: neutral)

-1.21*** (satisfied)

1.30*** (dissatisfied)

(ref: neutral)

0.37** (satisfied) -0.90 (satisfied)

-0.18 (fairly satisfied)

0.57 (dissatisfied)

(ref: satisfied)

Quality of equipment n/a n/a n/a n/a

-0.96** (fairly satisfied)

0.95** (dissatisfied)

(ref: satisfied)

Formal Health Costs 0.33 1.41*** 0.16 -0.84 -0.01

Informal Health Costs 0.28 0.87* n/a n/a 0.27

Provider of Health

Facilities 0.45** (government) 1.37*** (government)

-0.86 (not run by

government)

-0.84 (not run by

government)

-1.73** (private)

-0.76 (religious)

80

Explanatory

variable Nepal (neutral) Nepal (dissatisfied)

Pakistan

(Indifferent)

Pakistan (very

dissatisfied) Sri Lanka

Uganda

(dissatisfied)

-0.49 (NGO)

-19.87*** (other)

(ref: government)

Community Meeting

Held on Health 0.30 0.98** n/a n/a -0.29

Attended Community

Meeting on Health -0.17 -0.53 n/a n/a n/a

Information about

the regression

model

R2 0.53 0.53 0.65 0.65 n/a

Number of

observations 2973 2973 2062 2062 1407

81

Table 10: Comparative analysis on experience of education service: Satisfied.

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female respondent -0.02 -1.60** -0.51 (female

respondent) 0.45 -0.12

Age of respondent 0.03* -0.03 0.05 -0.06 0.00

Main activity of household -0.39 (main income

source: agriculture)

-0.52 (main income source: own

cultivation)

0.64 (main income:

farming)

-1.06 (main income:

overseas labour)

0.25 (main activity: fishing)

1.83 (main activity:

agriculture)

1.11 (main activity: trading)

Main income source:

0.88 (casual labour: agr)

-2.30** (casual labour: non-agr)

1.87** (exploitation of bush products)

8.63*** (other assistance)

-0.53 (other economic activity)

0.61 (own business: shop building)

-1.06 (business: home, market)

-2.38*** (own livestock)

-19.68*** (paid housework)

-0.96 (private sector, NGO)

0.45 (government)

(ref: none)

Education level of respondent -0.02 (mean years of

education) 0.26 (respondent’s education)

0.03 (Average education

of adults)

-0.88 (share of adults

completing primary)

Respondent education:

-0.18 (some primary)

-0.54 (primary)

-0.58 (O level)

0.84 (A level)

-0.43 (Tertiary)

(ref: none)

Household has migrant n/a -2.82**

-0.06 (internal migrant)

-0.02 (international

migrant)

-1.13** -3.08**

Household receives

remittances 0.74 1.83** 0.36 1.58 0.07

Household has been displaced -0.47 (displaced due to

conflict) -1.15 (lived in village all life)

0.05 (displaced due to

conflict) -2.93**(ever displaced) 0.44 (ever displaced)

Religion -0.26 (minority) n/a n/a n/a n/a

Ethnicity 0.04 (minority) 0.72 (minority) n/a 0.65 (minority)

-0.28 (Langi)

2.51** (Iteso)

-22.41*** (Kumam)

11.54*** (Karamojong)

-0.39 (Mixed)

-19.53*** (other)

(ref: Acholi)

Dependency ratio -0.14 -0.17 -0.29 -0.57 0.00

Morris Index 0.00 0.02 0.05 0.05 -0.10**

Food security index 0.04 0.13** 0.01 -0.02 -0.03*

82

Context

Urban n/a -1.56* n/a n/a 0.87

Household has experienced

conflict in past three years 0.27 0.98 0.76 n/a 0.48

Perception of safety

0.44 (safe in village)

-0.10 (safe while

travelling)

-0.53 (safe in neighbourhood) -0.41 (safe when moving

around) 1.12 (safe in neighbourhood) -0.62** (Safe in neighbourhood)

Location n/a

1.25 (Rolpa)

1.89** (Bardiya)

(ref: Ilam)

1.09** (Swat)

(ref: Lower Dir) n/a 0.62 (Sub-region: Langi) (reference: Acholi)

Shocks

Shocks experienced -0.13 0.03 -0.46** -0.24 -0.03

Crimes experienced -0.64*** 0.02 0.43 -1.04 -0.07** (crimes)

0.05 (serious crimes)

Perception of

education

Distance to school 0.00 0.00 -0.03 0.00 0.00

Number of teachers

-2.12*** (neutral)

-0.80 (dissatisfied)

(ref: satisfied)

-1.49 (satisfied: boys)

-0.52 (satisfied: girls)

1.41* (satisfied: boys)

1.09 (neutral: girls)

2.54* (satisfied: girls)

(ref: dissatisfied)

1.86* (satisfied: boys)

1.02 (satisfied: girls)

-0.93** (fairly satisfied)

-0.59 (dissatisfied)

(ref: satisfied)

Quality of teachers

-1.10* (neutral)

0.51 (dissatisfied)

(ref: satisfied)

3.96** (satisfied: boys)

-6.73*** (satisfied: girls) n/a n/a

-1.19*** (fairly satisfied)

-1.44 (dissatisfied)

(ref: satisfied)

Attendance of teachers

-0.61 (neutral)

-0.59 (dissatisfied)

(ref: satisfied)

-2.30* (satisfied: boys)

-0.72 (satisfied: girls)

-1.96 (neutral: boys)

-2.86 (satisfied: boys)

5.66** (neutral: girls)

6.94** (satisfied: girls)

(ref: dissatisfied)

n/a

-1.13*** (fairly satisfied)

-2.88*** (dissatisfied)

(ref: satisfied)

Size of class

-0.96** (dissatisfied)

-0.24 (satisfied)

(ref: neutral)

-1.17 (satisfied: boys)

0.11 (satisfied: girls)

1.39 (neutral: boys)

1.78* (satisfied: boys)

-0.39 (neutral: girls)

0.34 (satisfied: girls)

(ref: dissatisfied)

n/a

-1.50*** (fairly satisfied)

-1.77*** (dissatisfied)

(ref: satisfied)

School infrastructure

-1.24*** (dissatisfied)

-0.42 (satisfied)

(ref: neutral)

-1.03 (satisfied: boys)

1.60 (dissatisfied: boys)

-1.65 (satisfied: girls)

-0.29 (dissatisfied: girs)

(ref: neutral)

0.64 (neutral: boys)

-0.08 (satisfied: boys)

1.34* (neutral: girls)

3.41*** (satisfied: girls)

(ref: dissatisfied)

n/a

-1.87*** (fairly satisfied)

-1.89** (dissatisfied)

(ref: satisfied)

Quality of equipment

1.36*** (dissatisfied)

1.99*** (satisfied)

(ref: neutral)

-1.62 (satisfied: boys)

-1.96 (dissatisfied: boys)

-0.80 (satisfied: girls)

-1.04 (dissatisfied: girls)

(ref: neutral)

n/a -0.52 (satisfied: boys)

1.77 (satisfied: girls)

-0.05 (fairly satisfied)

-2.63*** (dissatisfied)

(ref: satisfied)

Official Fees for School -0.27 -0.50 (boys)

-2.32 (girls)

-1.18* (boys)

1.89** (girls)

13.49 (boys)

-12.59 (girls) n/a

Informal fees for school n/a -0.06 (boys)

0.67 (girls)

0.27 (boys)

-0.14 (girls)

2.73 (boys)

-1.44 (girls) n/a

Provider -0.38 (government) 2.82* (government: boys)

-4.89** (government: girls) n/a n/a

1.54** (private)

1.60 (religious)

83

40.01*** (NGO)

7.44*** (other)

(ref: government)

Community Meeting Held on

Education -0.35 0.00 0.87 1.01** 1.23***

Attended Community Meeting

on Education 0.10 -0.05 n/a n/a n/a

Information about

the regression

model

R2 0.44 0.66 0.53 0.43 n/a

Number of observations 400 463 503 306 774

Note: The outcome variable in this case is binary, corresponding to 1 if the respondent is satisfied with the overall quality of the boys’ and girls’ school. Since the model controls for aspects of boys’ and

girls’ schooling independently we only include households which have both boys and girls enrolled in primary school.

Table 11: Comparative analysis on experience of water services: Water is clean and safe.

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female respondent 0.23 0.22 -0.03 0.13 (male) 0.14 (female head)

Age of respondent -0.01 0.01** 0.03 0.02* 0.01

Main activity of household 0.10 (main income source:

agriculture)

-0.17 (main income source:

own cultivation)

0.14 (HH head in farming)

-0.19 (main income:

farming)

-0.22 (main income:

overseas labour)

-0.29 (respondent

employed)

Main income source:

1.81* (casual labour: agr)

0.25 (casual labour: non-

agr)

0.24 (exploitation of bush

products)

-1.36 (other assistance)

-0.14 (other economic

activity)

-1.06** (own business:

shop)

-0.11 (business: home,

market)

0.08 (own livestock)

0.34 (private/ NGO)

-0.09 (government)

Education level of

respondent 0.04 (categorical) 0.01 (categorical)

-0.03 (average education of

adults)

0.72 (respondent completed

primary)

Respondent educ:

-0.18 (some primary)

-0.18 (primary)

-0.30 (O level)

-0.43 (A level)

0.114 (Tertiary)

84

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

(ref: none)

Household has migrant n/a -0.17 0.12 (internal)

0.43 (international) 0.00 -0.05

Household receives

remittances 0.61* -0.22 -0.38 0.35 0.16

Household has been

displaced

-0.21 (displaced due to

conflict)

0.38**(lived in village all

life)

0.28 (displaced due to

conflict) -0.62 (ever displaced) 0.39** (ever displaced)

Religion 0.36 (minority) n/a n/a n/a n/a

Ethnicity 1.15*** (minority) 0.27 (minority) n/a -0.07 (minority)

0.58 (Langi)

0.43 (Kumam)

-0.80 (Karamojong)

1.56 (Mixed)

-0.20 (other)

(ref: Acholi)

Dependency ratio -0.16** -0.16* n/a 0.14 0.00

Morris Index 0.00 0.00 -0.00 -0.14** 0.04

Food Security Index 0.02 -0.05*** 0.00 -0.03 -0.01**

Context

Urban n/a -0.60*** n/a -0.02 0.53

Household has experienced

conflict in past three years 0.52** 0.11 -0.00* n/a 1.20

Perception of safety 0.15 (safe in village)

-0.03 (safe while travelling) 0.24 (safe in neighbourhood) 0.36* (safe moving around)

0.46 (safe in

neighbourhood)

-0.29**(safe in

neighbourhood)

Location n/a

-0.70*** (Rolpa)

-0.56*** (Bardiya)

(ref: Ilam)

-0.95*** (Swat)

(ref: Lower Dir) n/a

-0.16 (Sub-region: Langi)

(reference: Acholi)

Shocks

Shocks experienced -0.02 -0.02 -0.07 -0.05 -0.04**

Crimes experienced 0.18 -0.01 -0.16 -0.66*** 0.01 (crimes)

0.00 (serious crimes)

Perception of health

services

Have to Queue for Water -0.68*** -0.26** -0.85*** -1.33*** 0.33

Official Fees for Water -0.58** -0.58*** 0.00 -0.58** 1.28***

Who runs water

-0.92*** (Run by

Committee)

-0.64** (Run by NGO)

-0.22 (Run by Nobody)

(Ref: government)

0.85** (Run by government)

-0.09 (Run privately)

1.15** (Run by NGO)

(ref: Other)

n/a -0.10 (government)

2.48*** (government)

1.36*** (private)

1.14*** (community)

2.50*** (NGO)

(ref: nobody)

85

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Community Meeting Held on

Water -0.41 -0.15 -1.14 n/a 0.05

Attended community

Meeting on Water -0.21 0.28 -0.00 0.26 n/a

Information about the

regression model

R2 0.14 0.06 0.04 0.05 n/a

Number of observations 777 2901 2087 1377 1445

86

Table 12: Comparative analysis on experience of social protection.

Explanatory variable

DRC (food aid

has an impact)

Nepal (Transfer

helps a bit)

Nepal (Transfer

helps a lot)

Pakistan (Helps a

lot)

Sri Lanka (Samurdhi

transfer makes

some difference)

Uganda

Household factors

Female

respondent -0.14 0.87 0.73 0.54 0.50 (male)

(Not enough

observations)

Age of respondent -0.00 1.01 1.00 0.01 0.12*

Main activity of

household

0.10 (main income

source: agriculture)

1.11 (main income source:

own cultivation)

0.54 (main income

source: own

cultivation)

0.36 (HH head in

farming)

0.35 (main income:

farming)

1.6 (main income:

overseas labour)

-2.43 (respondent is

employed)

Education level of

respondent -0.08 (categorical) 1.10 (categorical) 1.25 (categorical)

-0.17*** (average

education of adults)

17.52 (respondent

completed primary)

Household has

migrant n/a 0.33** 1.19

-3.06 (internal)

-1.38 (international) -2.94**

Household

receives

remittances

-0.89 0.93 0.54 0.11 0.96

Household has

been displaced

0.58 (displaced due to

conflict) 0.73 (lived in village all life)

0.64 (lived in village

all life)

-1.87*** (displaced due

to conflict) 0.95 (ever displaced)

Number of children 0.1764* 1.00 0.83 0.19 (dependency ratio) 1.01

Number of elderly -0.06 0.98 1.78 n/a -2.35

Religion 0.77 (minority) n/a n/a n/a n/a

Ethnicity -0.78* (minority) 1.31 (minority) 1.27 (minority) n/a -19.63 (minority)

Morris Index 0.00 1.01 0.98 -0.03** -1.13*

Food Security

Index -0.00 0.98 1.04 0.06 -0.02

Context

Urban n/a 1.01 0.83 n/a 2.54

Household has

experienced

conflict in past

three years

-0.87 0.90 0.74 -0.28 n/a

87

Explanatory variable

DRC (food aid

has an impact)

Nepal (Transfer

helps a bit)

Nepal (Transfer

helps a lot)

Pakistan (Helps a

lot)

Sri Lanka (Samurdhi

transfer makes

some difference)

Uganda

Perception of

safety

-0.56 (safe in village)

0.69 (safe while

travelling)

2.85** (safe in

neighbourhood)

8.34* (safe in

neighbourhood)

0.36 (safe moving

around)

-2.76 (safe in

neighbourhood)

Location n/a

1.37 (Rolpa)

0.50** (Bardiya)

(ref: Ilam)

3.12* (Rolpa)

0.66 (Bardiya)

(ref: Ilam)

-2.89*** (Swat) n/a

Shocks

Shocks

experienced 0.01 0.91** 0.85 -0.01 -1.38*

Crimes

experienced 0.07 1.42** 1.65*** 0.61 -15.12

Perception of

transfer

Right amount n/a 0.56 (not always right

amount)

3.69 (not always the

right amount) 0.11 (right amount) 0.15 (right amount)

Received on Time -3.39*** (on time)

1.16 (sometimes)

0.02*** (never)

(ref: always)

0.63 (sometimes)

0.25 (never)

(ref: always)

-0.28 (on time) 1.13 (one time)

Provider -1.30** (not an NGO) n/a n/a n/a n/a

Community

Meeting Held

(Social Protection)

n/a 0.60 0.00 6.57*** n/a

Attended Meeting

on Social

Protection

n/a 1.51 0.91 5828.64 0.13

Information about

the regression model

R2 0.25 0.16 0.16 0.13 0.11

Number of

observations 161 376 376 400 255

88

Table 13: Comparative analysis on experience of livelihood services.

Explanatory

variable DRC

Nepal (seeds and

grains transfer

improved

production)

Pakistan (seeds and

tools transfer improved

production)

Sri Lanka (Impact of

livelihood service on

production)

Uganda (has an

impact)

Household factors

Female respondent (Not enough

observations of transfer

receivers to carry out

regression)

0.23 n/a 0.39 (male) -1.06 (female head)

Age of respondent 0.00 0.02 (average age of HH) -0.01 0.01

Main activity of household 0.96* (main income source:

own cultivation)

0.04 (number of activities)

-1.84 (ratio of self-employed)

0.62 (ratio of unemployed)

0.07 (respondent is

employed)

Main income source:

0.02 (business: home,

market)

0.94 (livestock)

Education level of

respondent -0.40** (categorical) 0.03 (categorical)

-0.22 (respondents

completed primary)

Respondent educ:

0.19 (some primary)

0.10 (primary)

-0.12 (O level)

-0.11 (tertiary)

(ref: none)

Household has migrant 0.93 -0.03 0.95** -0.97

Household receives

remittances 0.36 n/a -0.12 -0.30

Household has been

displaced 0.28 (lived in village all life) -0.68 (displaced due to conflict) 0.68 (ever displaced) 0.22 (ever displaced)

Dependency ratio -0.28 n/a -0.18 0.00

Household owns land 3.50 1.16 0.00 -0.01

Ethnicity -0.47 (minority) n/a -0.80* (minority)

-0.92 (Langi)

-0.02 (Mixed)

(ref: Acholi)

Morris Index 0.05** -0.00 0.02 0.14**

Food Security Index -0.12** 0.03 -0.05* -0.04**

Context

Urban 1.21 n/a -0.52 0.90

Household has experienced

conflict in past three years 0.20 n/a n/a n/a

Perception of safety 1.36* (safe in

neighbourhood) n/a 0.53 (safe in neighbourhood) -0.44 (safe in neighbourhood)

Location 0.44 (Rolpa) n/a n/a 3.15 (Sub-region: Langi)

89

Explanatory

variable DRC

Nepal (seeds and

grains transfer

improved

production)

Pakistan (seeds and

tools transfer improved

production)

Sri Lanka (Impact of

livelihood service on

production)

Uganda (has an

impact)

0.68 (Bardiya)

(ref: Ilam)

(reference: Acholi)

Shocks

Shocks experienced -0.08 -0.41* 0.05 0.01

Crimes experienced -0.03 -0.09 -0.35 0.03 (crimes)

0.01 (serious crimes)

Barriers to agriculture n/a 0.21 n/a n/a

Perception

of transfer

Received on time 2.01*** (on time) n/a 3.14*** (on time) 2.74*** (on time)

Who provides service

-0.48 (Government)

0.55 (International NGO)

(ref: Other)

3.33*** (non-government) 0.52 (government)

0.85* (national NGO)

0.49 (international NGO)

0.95 (community)

-0.61 (religious)

(ref: government)

Community Meeting Held

(Livelihood Services) -0.92 n/a n/a 0.05

Attended Meeting on

Livelihood -0.01 n/a 1.12** n/a

Information about

the regression

model

R2 0.29 0.41 0.27 n/a

Number of observations 195 2108 446 214

Table 14: Comparative analysis: Local government cares about opinions

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household

factors

Female respondent -0.22 -0.12 0.23 -0.47 (male) -0.26*

Age of respondent 0.03 0.00 0.00 -0.01 -0.01**

Main activity of household 1.03 (main income source:

agriculture)

0.02 (main income source: own

cultivation)

0.19 (farming main

activity of head)

1.03*** (casual labour

main activity of head)

0.80** (casual labour,

non-agri, main activity of

head)

-0.52 (own business main

activity of head)

0.13 (respondent is

employed)

Respondent livelihood:

-0.09 (casual labour: agr)

-0.14 (casual labour: non-agr)

-0.14 (exploitation of bush products)

-0.06 (no activity)

0.02 (other economic activity)

-0.41 (own business: shop building)

0.56* (business: home, market)

-0.77 (fishing)

90

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

-1.21*** (head has no

activity)

-0.36 (livestock)

0.56 (paid housework)

-0.25 (private/ NGO)

0.01 (work for government)

(ref: none)

Education level of respondent 0.14 (categorical) 0.06* (categorical) -0.42** (head has some

education)

0.73* (completed

primary)

Respondent educ:

0.17 (some primary)

0.16 (primary)

-0.09 (O level)

-0.28 (A level)

-0.07 (tertiary)

(ref: none)

Household has migrant n/a 0.16 0.15 -0.12 0.31

Household receives remittances 0.87 0.34*** -0.08 0.40 0.26

Household has been displaced -0.23 (displaced due to

conflict)

0.01 (lived in village whole life) -0.03 (displaced due to

conflict)

-0.20 (ever displaced) -0.20 (ever displaced)

Dependency ratio -0.03 0.03 -0.35** 0.05 0.00

Religion 0.46 (minority) n/a n/a n/a n/a

Ethnicity -0.25 (minority) -0.16 (minority) n/a 0.16 (minority)

-0.17 (Langi)

0.15 (Kumam)

-1.37** (Mixed)

-0.16 (other)

(ref: Acholi)

Morris Index 0.00 0.00** 0.00 -0.04 0.01

Food Security Index 0.03 -0.01 -0.04 -0.02 0.00

Context

Urban n/a 0.02 n/a -0.10 -0.10

Household has experienced conflict

in past three years 0.64 -0.28** n/a n/a 0.23

Perception of safety

0.31 (safe in village)

-0.76 (safe while

travelling)

0.02 (safe in neighbourhood) 1.20*** (safe when

moving around)

1.19** (safe in

neighbourhood) -0.24*** (safe in neighbourhood)

Location n/a

0.07 (Rolpa)

-0.69*** (Bardiya)

(ref: Ilam)

0.84** (Swat) (ref: Lower

Dir) n/a

-0.47*(Sub-region: Langi) (reference:

Alcholi)

Shocks

Shocks experienced 0.24 0.02 0.19 -0.07 0.04**

Crimes experienced -0.23 -0.05** -0.09 -0.33* 0.00 (crimes)

-0.01 (serious crimes)

Access to Distance to health -0.0168* 0.00 0.00 0.00 0.00**

91

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

services Distance to water 0.00 -0.01* n/a -0.01** 0.00

Receives social protection 0.42 0.03 0.18 -0.12 0.30

Receives livelihood transfer -0.35 0.20 -0.07 0.03 0.05

Perception of

services

Satisfied with health -0.98 (satisfied) 0.05 (Dissatisfied)

-0.06 (satisfied)

0.38** (satisfied)

-0.47* (fairly satisfied)

-0.87*** (dissatisfied)

(ref: satisfied)

Number of staff at health facility

1.34* (neutral)

-0.60 (satisfied)

(ref: dissatisfied)

0.14 (satisfied)

-0.16 (dissatisfied)

(ref: neutral)

-0.48 (neutral)

1.01* (satisfied)

(ref: dissatisfied)

n/a

-0.23 (fairly satisfied)

-0.41 (dissatisfied)

(ref: satisfied)

Availability of equipment/ medicine

-0.30 (neutral)

-0.17 (satisfied)

(ref: dissatisfied)

0.28* (satisfied)

-0.05 (dissatisfied)

(ref: neutral)

-0.24 (neutral)

-0.36 (satisfied)

(ref: dissatisfied)

0.32** (satisfied)

0.10 (fairly satisfied with equipment)

0.34 (dissatisfied with equipment)

(ref: satisfied)

Waiting time at health facility

0.63 (neutral)

2.02*** (satisfied)

(ref: dissatisfied)

-0.40*** (satisfied)

0.11 (dissatisfied)

(ref: neutral)

0.13 (neutral)

-0.26 (satisfied)

(ref: dissatisfied)

-0.05 (satisfied)

-0.41 (fairly satisfied)

-0.64 (dissatisfied)

(ref: satisfied)

Queue for water 0.31 0.17 -1.42*** -0.11 0.24

Water clean and safe 0.82 -0.15 0.64 0.20 0.01

Formal fees for health -0.11 -0.17 -0.41 0.48 -0.35

Informal fees for health -1.40* -0.54** n/a -0.02 0.37*

Fees for water 1.44** 0.03 n/a -0.87*** 0.21

Provider of health service 0.21 (Government) 0.29** (Government) -0.42 (not government) 0.41 (Government)

-0.31 (private)

-0.29 (religious)

-1.20 (NGO)

(ref: government)

Provider of water

-0.32 (Committee)

-0.05 (NGO)

-0.87 (Nobody)

(Ref: government)

0.41* (Government)

0.51** (Private owner)

0.54* (NGO)

(ref: other)

0.62* (not government) -0.16 (Government)

0.43 (government)

0.38 (private)

0.44** (community)

0.68* (NGO)

-0.17 (other)

(ref: nobody)

Community

involvement

Experienced problem with service -0.13 -0.20*** -0.04 -0.16** -0.16*** (number of problems)

Know of way to make a complaint n/a 0.14*** 1.24*** 0.01 0.35* (fraction of services in which a

complaint was made)

92

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Attended meeting on any service

0.01 (Know of meeting on

health)

0.32 (Attended meeting on

health)

2.47** (Know of meeting

on water)

-0.76 (Attended meeting

on water)

0.00 (Know of meeting)

0.13**(Attended meeting)

-0.72 (know of any

meetings)

0.97 (number of meetings

attended)

0.53*** (knew of

meetings)

-0.02 (number of

meetings attended)

0.15** (number of meetings)

Was consulted around any service n/a 0.11** 0.81*** 0.17** 0.19** (number of consultations)

Information

about the

regression

model

R2 0.28 0.10 n/a 0.18 0.03

Number of observations 189 2517 2064 1201 1377

Table 15: Comparative analysis: Local government decisions reflect respondent’s priorities (Never or almost never).

Explanatory

variable

DRC (Only in some

areas)

Nepal (Never or

almost never)

Pakistan (Only in some

areas) Sri Lanka Uganda (Almost never)

Household factors

Female respondent -0.49 -0.02 -0.19 0.35* (male) 0.13

Age of respondent 0.01 -0.01 -0.02** 0.02*** 0.00

Main activity of

household

-0.77** (main income

source: agriculture)

-0.2 (main income source:

own cultivation)

1.12** (farming main activity of

head)

1.67*** (casual labour main

activity of head)

1.69*** (casual labour, non-agri,

main activity of head)

0.89 (own business main activity of

head)

0.15 (head has no activity)

-0.27 (respondent is

employed)

Respondent livelihood:

0.11 (casual labour: agr)

1.28*** (casual labour: non-agr)

0.71* (exploitation of bush

products)

0.11 (no activity)

1.20*** (other economic activity)

0.71 (own business: shop building)

0.21 (business: home, market)

-20.44*** (fishing)

0.07 (livestock)

0.21 (paid housework)

1.08 (private/ NGO)

-0.16 (work for government)

(ref: none)

Education level of

respondent 0.00 (categorical) -0.13 (categorical) -0.37 (head has some education)

-0.21 (respondent

completed primary)

Respondent educ:

0.11 (some primary)

0.47 (primary)

0.67* (O level)

1.55* (A level)

0.84* (tertiary)

(ref: none)

93

Explanatory

variable

DRC (Only in some

areas)

Nepal (Never or

almost never)

Pakistan (Only in some

areas) Sri Lanka Uganda (Almost never)

Household has migrant n/a -0.65 -0.28 0.36* -0.29

Household receives

remittances 0.26 0.09 0.43 -0.12 0.04

Household has been

displaced

0.06 (displaced due to

conflict)

0.01 (Lived in village whole

life) -0.12 (displaced due to conflict)

-0.21 (ever

displaced) 0.09 (ever displaced)

Dependency ratio -0.04 0.02 -0.28 -0.08 0.00

Religion 0.49 (minority) n/a n/a n/a n/a

Ethnicity -0.69 (minority) -0.49* (minority) n/a -0.92*** (minority)

-0.24 (Langi)

-0.54 (Iteso)

-0.19 (Kumam)

-22.04*** (Karamojong)

-0.50 (Mixed)

-1.99 (other)

(ref: Acholi)

Morris Index 0.00 0.00 0.00 0.11 -0.03*

Food Security Index 0.06** -0.01 -0.06* 0.01 0.00

Context

Urban n/a 0.05 n/a 0.18 -0.48

Household has

experienced conflict in

past three years

-0.44 0.03 n/a n/a -0.10

Perception of safety 0.78*(safe in village)

-0.34 (safe while travelling)

-0.08 (safe in

neighbourhood) 0.15 (safe when moving around)

-0.96** (safe in

neighbourhood) -0.26** (safe in neighbourhood)

Location n/a

0.44 (Rolpa)

0.05 (Bardya)

(ref: Ilam)

1.46*** (Swat) (ref: Lower Dir) n/a 0.15 (Sub-region: Langi) (reference:

Alcholi)

Shocks

Shocks experienced 0.02 0.00 -0.09 0.18** 0.05**

Crimes experienced -0.30 -0.03 -0.13 0.09 0.00 (crimes)

-0.01 (serious crimes)

Access to services

Distance to health 0.00 0.00 -0.01** 0.00 0.00***

Distance to water 0.00 0.00 n/a 0.01* 0.00

Receives social

protection 0.41 0.13 0.21 -0.10 -0.20

Receives livelihood

transfer -0.20 0.05 -0.29 -0.55*** 0.27

94

Explanatory

variable

DRC (Only in some

areas)

Nepal (Never or

almost never)

Pakistan (Only in some

areas) Sri Lanka Uganda (Almost never)

Perception of

services

Satisfied with health

service overall 0.53* (satisfied) 1.06*** (Dissatisfied) -1.00** (satisfied)

-0.22 (satisfied)

-0.02 (fairly satisfied)

-0.91* (dissatisfied)

(ref: satisfied)

Number of staff at

health facility n/a

0.41 (satisfied)

1.22** (dissatisfied)

(ref: neutral)

-2.15*** (neutral)

-0.53 (satisfied)

(ref: dissatisfied)

n/a

0.14 (fairly satisfied)

-0.29 (dissatisfied)

(ref: satisfied)

Availability of

equipment/ medicine n/a

0.55* (satisfied)

0.36 (dissatisfied)

(ref: neutral)

0.26 (neutral)

0.96* (satisfied)

(ref: dissatisfied)

-0.45 (satisfied)

0.22 (fairly satisfied with equipment)

-0.49 (dissatisfied with equipment)

(ref: satisfied)

Waiting time at health

facility n/a

-0.34 (satisfied)

-0.39 (dissatisfied)

(ref: neutral)

-0.48 (neutral)

-0.49 (satisfied)

(ref: dissatisfied)

0.12 (satisfied)

0.01 (fairly satisfied)

0.21 (dissatisfied)

(ref: satisfied)

Queue for water -0.21 0.49** -0.99** 0.16 0.22

Water clean and safe 0.43 -0.06 0.36 -0.57** 0.29

Formal fees for health 12.67 -0.38 n/a 0.68 0.12

Informal fees for health -0.37 0.39 n/a -0.06 0.14

Fees for water 0.28 -0.55* n/a 0.87*** 0.52***

Provider of health

service 0.12 (Government) -0.16 (Government) n/a 0.82* (Government)

-0.08 (private)

-0.72 (religious)

-1.84 (NGO)

-22.55 (other)

(ref: government)

Provider of water

0.12 (Committee)

0.20 (NGO)

-0.87 (Nobody)

(Ref: government)

0.71 (Government)

0.37 (Private)

-1.63 (NGO)

(ref: other)

n/a -0.36 (Government)

0.21 (government)

0.15 (private)

0.34 (community)

0.00 (NGO)

0.52 (other)

(re: nobody)

Community

involvement

Experienced problem

with service -0.03 -0.06 -0.11 0.06 -0.05 (number of problems)

Know of way to make a

complaint n/a 0.07 0.95** -0.10

-0.12 (fraction of services in which a

complaint was made)

Attended meeting on

any service n/a

0.00 (Know of a meeting)

0.12 (Attended a meeting)

0.00 (know of any meetings)

0.27 (number of meetings attended)

-0.34 (knew of

meetings) 0.16* (number of meetings)

95

Explanatory

variable

DRC (Only in some

areas)

Nepal (Never or

almost never)

Pakistan (Only in some

areas) Sri Lanka Uganda (Almost never)

0.15 (number of

meetings attended)

Was consulted around

any service n/a 0.16 -11.42 -0.09 -0.26** (number of consultations)

Information

about the

regression

model

R2 0.13 0.10 n/a 0.28 n/a

Number of

observations 502 1417 2088 1303 1373

Table 16: Comparative analysis: Local government decisions reflect respondent’s priorities (Very much/ to a large extent or Always).

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan

(Always)

Sri Lanka

(Completely)

Uganda (Absolutely/

always)

Household

factors

Female respondent -0.51 -0.3 0.43 -0.29 (male) -0.12

Age of respondent 0.02 0.02 0.01 0.00 -0.02*

Main activity of household -0.81** (main income

source: agriculture)

0.48 (main income source:

own cultivation)

-0.32 (farming main

activity of head)

0.47(casual labour

main activity of head)

0.24 (casual labour,

non-agri, main activity

of head)

-0.54 (own business

main activity of head)

-1.23*** (head has

no activity)

-0.04 (respondent is

employed)

Respondent livelihood:

0.03 (casual labour: agr)

0.09 (casual labour: non-agr)

0.01 (exploitation of bush products)

0.31 (no activity)

0.54 (other economic activity)

0.36 (own business: shop building)

0.71 (business: home, market)

0.15 (fishing)

0.09 (livestock)

0.88 (paid housework)

0.79 (private/ NGO)

1.28** (work for government)

(ref: none)

Education level of respondent -0.04 (categorical) 0.24** (categorical) -0.27 (head has some

education)

-0.42 (respondent completed

primary)

Respondent educ:

0.2 (some primary)

0.36 (primary)

-0.21 (O level)

0.72 (A level)

-0.21 (tertiary)

(ref: none)

Household has migrant n/a -0.32 -0.23 -0.09 0.17

96

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan

(Always)

Sri Lanka

(Completely)

Uganda (Absolutely/

always)

Household receives

remittances -0.10 0.43 -0.24 0.10 -0.24

Household has been

displaced

-0.77 (displaced due to

conflict)

-0.1 (Lived in village whole

life)

-0.40 (displaced due to

conflict) -0.26 (ever displaced) 0.01 (ever displaced)

Dependency ratio 0.29** -0.35 -0.12 0.02 0.00

Religion -0.67 (minority) n/a n/a n/a n/a

Ethnicity 0.59 (minority) 0.31 (minority) n/a -0.72*** (minority)

0.59 (Langi)

-21.22*** (Iteso)

-0.03 (Kumam)

-21.47*** (Karamojong)

-0.18 (Mixed)

0.43 (other)

(ref: Acholi)

Morris Index 0.00 0 0.01* 0.01 0.00

Food Security Index 0.08*** -0.02 -0.01 -0.03** -0.02**

Context

Urban n/a 0.2 n/a 0.42** -0.25

Household has experienced

conflict in past three years -0.92** -0.14 n/a n/a 0.32

Perception of safety

0.33 (safe in village)

-0.94** (safe while

travelling)

-1.05** (safe in

neighbourhood)

1.6** (safe when

moving around) 0.90 (safe in neighbourhood) -0.33** (safe in neighbourhood)

Location n/a

0.94 (Rolpa)

1.10** (Bardya)

(ref: Ilam)

-0.08 (Swat) (ref:

Lower Dir) n/a

-0.9* (Sub-region: Langi) (reference:

Alcholi)

Shocks

Shocks experienced 0.34*** 0.10*** 0.45** -0.02 0.05*

Crimes experienced -0.47** -0.14 -0.06 -0.02 -0.01 (crimes)

-0.05** (serious crimes)

Access to services

Distance to health -0.01 -0.01 0.00 0.00 0.00

Distance to water 0.01 -0.01 n/a 0.00 0.00

Receives social protection 1.01** 0.01 0.24 -0.12 0.53

Receives livelihood transfer -0.22 0.59 0.62 -0.29* 0.06

Perception of Satisfied with health service

overall 0.06 (satisfied) 0.16 (dissatisfied) -0.58 (satisfied) 0.60*** (satisfied)

-0.63 (fairly satisfied)

-0.99** (dissatisfied)

97

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan

(Always)

Sri Lanka

(Completely)

Uganda (Absolutely/

always)

services (ref: satisfied)

Number of staff at health

facility n/a

-0.51 (satisfied)

-0.37 (dissatisfied)

(ref: neutral)

0.06 (neutral)

0.80 (satisfied)

(ref: dissatisfied)

n/a

-0.39 (fairly satisfied)

-0.68 (dissatisfied)

(ref: satisfied)

Availability of equipment/

medicine n/a

1.95*** (satisfied)

1.85** (dissatisfied)

(ref: neutral)

-0.50 (neutral)

-0.65 (satisfied)

(ref: dissatisfied)

-0.10 (satisfied)

0.41 (fairly satisfied with

equipment)

0.38 (dissatisfied with equipment)

(ref: satisfied)

Waiting time at health facility n/a

-0.46 (satisfied)

-0.34 (dissatisfied)

(ref: neutral)

0.19 (neutral)

0.13 (satisfied)

(ref: dissatisfied)

0.60*** (satisfied)

-0.81* (fairly satisfied)

-0.74 (dissatisfied)

(ref: satisfied)

Queue for water 0.27 0.15 -0.85 -0.41 0.17

Water clean and safe 0.90** -0.94** 0.59 -0.16 0.08

Formal fees for health 13.31 -0.33 n/a 0.09 -0.30

Informal fees for health -0.37 1.15** n/a 0.07 -0.01

Fees for water -0.13 -0.02 n/a -0.48** 0.14

Provider of health service -0.50 (Government) 0.71* (Government) n/a 0.44

0.07 (private)

-0.41 (religious)

-0.76 (NGO)

-21.49*** (other)

(ref: government)

Provider of water

-0.02 (Committee)

0.22 (NGO)

-0.79 (Nobody)

(Ref: government)

-0.7 (Government)

-0.92 (Private)

-1.52 (NGO)

(ref: other)

n/a -0.08

0.73*** (government)

0.07 (private)

0.32 (community)

0.58 (NGO)

-0.32 (other)

(ref: nobody)

Community

involvement

Experienced problem with

service -0.08 -0.90*** -0.37** -0.33*** -0.15* (number of problems)

Know of way to make a

complaint n/a 0.28*** 1.18** 0.21

-0.54 (fraction of services in which a

complaint was made)

Attended meeting on any

service n/a

0.00 (Knew of meeting)

-0.10 (Attended meeting)

-064 (know of any

meetings)

1.13 (number of

meetings attended)

0.61*** (knew of meetings)

-0.31 (number of meetings

attended)

0.39*** (number of meetings)

98

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan

(Always)

Sri Lanka

(Completely)

Uganda (Absolutely/

always)

Was consulted around any

service n/a 0.32** 0.51 0.54*** 0.18* (number of consultations)

Information about

the regression

model

R2 0.13 0.10 n/a 0.28 n/a

Number of observations 502 1417 2088 1303 1373

Table 17: Comparative analysis: Central government cares about opinions.

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

Household factors

Female respondent 0.01 -0.07 -0.20 -0.33* (male) -0.03

Age of respondent 0.02 0.00 -0.01 0.00 -0.01*

Main activity of

household

0.60 (main income

source: agriculture)

-0.29** (main income source:

own cultivation)

-0.11 (farming main activity of

head)

-0.09 (casual labour main

activity of head)

-0.02 (casual labour, non-agri,

main activity of head)

-0.62(own business main

activity of head)

-1.31*** (head has no activity)

0.22 (respondent is

employed)

Respondent livelihood:

-0.45* (casual labour: agr)

-0.99*** (casual labour: non-agr)

-0.97*** (exploitation of bush

products)

-0.43 (no activity)

(other economic activity)

-0.31 (own business: shop building)

0.51 (business: home, market)

-1.3 (fishing)

-0.78*** (livestock)

-0.46 (paid housework)

-0.84 (private/ NGO)

-0.52* (work for government)

(ref: none)

Education level of

respondent

0.41***

(categorical) 0.08** (caegorical)

-0.09 (head has some

education)

0.20 (respondent

completed primary)

Respondent educ:

-0.01 (some primary)

0.22 (primary)

-0.35 (O level)

-0.19 (A level)

-0.25 (tertiary)

(ref: none)

Household has migrant n/a 0.47** -0.21 0.05 0.22

Household receives

remittances 0.21 0.12 -0.18 0.21 0.20

Household has been -0.86 (displaced due 0.23* (Lived in village whole 0.88 (displaced due to conflict) -0.46 (ever displaced) -0.19 (ever displaced)

99

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

displaced to conflict) life)

Dependency ratio 0.33* -0.08 -0.32 0.33** 0.00

Religion 0.10 (minority) n/a n/a n/a n/a

Ethnicity 0.58 (minority) n/a n/a -0.28 (minority)

-0.11 (Langi)

0.63 (Kumam)

-0.53 (Mixed)

0.77 (other)

(ref: Acholi)

Morris Index 0.00 0.00 0.00 0.06 0.00

Food Security Index 0.06 0.01 -0.09** 0.05*** -0.01

Context

Urban n/a 0.33** n/a -0.03 -0.28

Household has

experienced conflict in

past three years

0.47 -0.13 n/a n/a -0.03

Perception of safety

-0.39 (safe in village)

-0.85 (safe while

travelling)

0.30 (Safe in neighbourhood) 0.22 (safe when moving

around)

0.69 (safe in

neighbourhood) -0.13 (safe in neighbourhood)

Location n/a

0.57*** (Rolpa)

-1.00*** (Bardiya)

(ref: Ilam)

0.66* (Swat) (ref: Lower Dir) n/a -0.15 (Sub-region: Langi) (reference:

Alcholi)

Shocks

Shocks experienced 0.18 -0.02 0.20 -0.15** 0.04**

Crimes experienced -0.31 -0.01 0.14 -0.40 -0.02 (crimes)

-0.02 (serious crimes)

Access to services

Distance to health 0.00 0.00 0.00 0.00 0.00

Distance to water -0.01 0.00 n/a -0.01 0.00

Receives social

protection -0.14 0.19 0.49* -0.22 -0.15

Receives livelihood

transfer 0.21 0.18 -0.44 0.38** 0.08

Perception

of services

Satisfied with health

service overall 0.60 (satisfied) 0.11 (Dissatisfied) -0.71 (satisfied) 0.27 (satisfied)

-0.52** (fairly satisfied)

-0.83** (dissatisfied)

(ref: satisfied)

Number of staff at health

facility

0.68 (neutral)

0.10 (satisfied)

(ref: dissatisfied)

0.06 (satisfied)

-0.18 (dissatisfied)

0.41 (neutral)

1.41** (satisfied)

(ref: dissatisfied)

n/a

-0.56*** (fairly satisfied)

-0.49** (dissatisfied)

100

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

(ref: neutral) (ref: satisfied)

Availability of equipment/

medicine

-0.60 (neutral)

-0.24 (satisfied)

(ref: dissatisfied)

0.32** (satisfied)

0.03 (dissatisfied)

(ref: neutral)

-0.37 (neutral)

-0.85** (satisfied)

(ref: dissatisfied)

0.23 (satisfied)

0.18 (fairly satisfied with equipment)

0.59** (dissatisfied with equipment)

(ref: satisfied)

Waiting time at health

facility

1.97** (neutral)

1.68* (satisfied)

(ref: dissatisfied)

-0.27(satisfied)

0.01 (dissatisfied)

(ref: neutral)

0.64 (neutral)

0.65 (satisfied)

(ref: dissatisfied)

-0.31* (satisfied)

-0.12 (fairly satisfied)

-0.66** (waiting time)

Queue for water 0.28 -0.13 -0.75* -0.10 0.33**

Water clean and safe 1.50** 0.08 0.19 -0.40 -0.12

Formal fees for health n/a -0.43** -0.26 -0.46 -0.14

Informal fees for health -1.75* -0.21 n/a 0.42 0.24

Fees for water -0.25 0.07 n/a -0.99*** 0.17

Provider of health service -0.38 (Government) 0.17 (Government) -0.45 (not government) -0.84 (Government)

-0.31 (private)

-0.56 (religious)

-1.16 (NGO)

(ref: government)

Provider of water

-0.84 (Committee)

-0.19 (NGO)

-1.07 (Nobody)

(Ref: government)

0.83** (Government)

0.80** (Private owner)

0.86** (NGO)

(ref: Other)

0.38 (not government) -0.02 (Government)

0.37 (government)

-0.22 (private)

0.12 (community)

0.39 (NGO)

0.05 (other)

(ref: nobody)

Community

involvement

Experienced problem

with service -0.22 -0.15** 0.07 -0.32*** -0.10*(number of problems)

Know of way to make a

complaint n/a 0.13*** 1.13*** 0.19

0.71*** (fraction of services in which

a complaint was made)

Attended meeting on any

service

-0.48 (Know of

meeting on health)

0.01 (Attended

meeting on health)

1.72 (Know of

meeting on water)

-1.05 (Attended

meeting on water)

0.00 (Know of meeting)

0.13**(Attended meeting)

-0.13 (know of any meetings)

0.40 (number of meetings

attended)

-0.25 (knew of meetings)

0.43** (number of

meetings attended)

0.06 (number of meetings attended)

Was consulted around n/a 0.09* 0.74** 0.07 0.18** (number of consultations)

101

Explanatory variable DRC Nepal Pakistan Sri Lanka Uganda

any service

Information about

the regression model

R2 0.19 0.13 n/a 0.14 n/a

Number of observations 181 2393 2064 1016 1368

Table 18: Comparative analysis: Central government decisions reflect respondent’s priorities (Never or almost never).

Explanatory variable

DRC (Only in

certain areas)

Nepal (Never/ Almost

never) Pakistan Sri Lanka

Uganda (Almost

never)

Household factors

Female respondent -0.42 0.21 -0.02 0.12 (male) 0.38**

Age of respondent 0.01 0.01 -0.01 0.00 0.01

Main activity of household -0.13 (main income

source: agriculture)

-0.36* (main income source:

own cultivation)

-0.62 (farming main

activity of head)

-1.02 (casual labour

main activity of head)

0.16 (casual labour,

non-agri, main activity

of head)

0.13 (own business

main activity of head)

-1.37*** (head has no

activity)

0.15 (respondent is employed)

Respondent livelihood:

-0.44 (casual labour: agr)

0. 54 (casual labour: non-agr)

-0.13 (exploitation of bush

products)

-0.95 (no activity)

0.63** (other economic

activity)

0.94** (own business: shop

building)

-0.78 (business: home, market)

-0.60 (fishing)

-0.22 (livestock)

-0.43 (paid housework)

-0.15 (private/ NGO)

0.68* (work for government)

(ref: none)

Education level of

respondent -0.02 (categorical) 0.06 (categorical)

-0.07 (head has some

education)

0.48 (respondent completed

primary)

Respondent educ:

0.55** (some primary)

0.41 (primary)

0.92** (O level)

0.85 (A level)

0.81** (tertiary)

(ref: none)

Household has migrant n/a 0.28 -1.07* -0.05 -0.15

Household receives

remittances 0.86 0.2 1.00* 0.19 -0.15

Household has been 0.63 (displaced due to 0.13 (Lived in village whole 0.59 (displaced due to 0.39 (ever displaced) 0.36 (ever displaced)

102

Explanatory variable

DRC (Only in

certain areas)

Nepal (Never/ Almost

never) Pakistan Sri Lanka

Uganda (Almost

never)

displaced conflict) life) conflict)

Dependency ratio 0.16 -0.01 0.05 -0.19 0.00**

Religion 0.47 (minority) n/a n/a n/a n/a

Ethnicity -0.80 (minority) -0.39 (minority) n/a -0.13 (minority)

0.13 (Langi)

-24.03*** (Iteso)

0.11 (Kumam)

-22.24*** (Karamojong)

-1.41 (Mixed)

-0.96 (other)

(ref: Acholi)

Morris Index 0.01* 0.00 0.00 -0.02 -0.05***

Food Security Index 0.02 0.03* -0.11** -0.05*** 0.01

Context

Urban n/a -0.43 n/a 0.40** 0.47

Household has

experienced conflict in

past three years

-0.12 0.33 n/a n/a -23.18***

Perception of safety 0.13 (safe in village)

0.17 (safe while travelling) 0.18 (safe in neighbourhood)

0.23(safe when moving

around) -1.35** (safe in neighbourhood)

-0.31**(safe in

neighbourhood)

Location n/a

0.68**(Rolpa)

0.69**(Bardiya)

(ref: Ilam)

0.35 (Swat) (ref: Lower

Dir) n/a

-0.05 (Sub-region: Langi)

(reference: Alcholi)

Shocks

Shocks experienced -0.20** 0.00 0.09 0.31*** 0.03

Crimes experienced -0.26 -0.03 -0.01 -0.14 0.05** (crimes)

-0.10*** (serious crimes)

Access to services

Distance to health 0.01 0.00 -0.01 0.00 0.00**

Distance to water 0.00 -0.01 n/a 0.02** 0.00*

Receives social protection 0.33 -0.17 0.06 0.21 -0.13

Receives livelihood

transfer 0.50 -0.1 -0.31 -0.47** -0.03

Perception of

services

Satisfied with health

service overall 0.28 (satisfied) 0.82** (Dissatisfied) n/a 0.04 (satisfied)

-0.01 (fairly satisfied)

-0.58 (dissatisfied)

(ref: satisfied)

Number of staff at health

facility n/a

-0.15 (satisfied)

n/a n/a

-0.11 (fairly satisfied)

103

Explanatory variable

DRC (Only in

certain areas)

Nepal (Never/ Almost

never) Pakistan Sri Lanka

Uganda (Almost

never)

-0.60 (dissatisfied)

(ref: neutral)

-0.96** (dissatisfied)

(ref: satisfied)

Availability of equipment/

medicine n/a

0.6** (satisfied)

0.09 (dissatisfied)

(ref: neutral)

n/a -0.27 (satisfied)

0.35 (fairly satisfied with

equipment)

0.18 (dissatisfied with

equipment)

(ref: satisfied)

Waiting time at health

facility n/a

-0.17 (satisfied)

-0.06 (dissatisfied)

(ref: neutral)

n/a -0.16 (satisfied)

-0.31 (fairly satisfied)

-0. 34 (dissatisfied)

(ref: satisfied)

Queue for water 0.14 0.18 n/a 0.35 -0.17

Water clean and safe 0.18 -0.03 n/a -0.11 0.26

Formal fees for health 12.52 n/a n/a 0.97** 0.36

Informal fees for health -0.66 0.71** n/a 0.45 0.21

Fees for water 0.54 -0.18 n/a 0.95** 0.37**

Provider of health service 0.04 (Government) 0.05 (Government) n/a 1.32*** (Government)

-0.74* (private)

-0.72 (religious)

-1.16 (NGO)

(ref: government)

Provider of water

-0.22 (Committee)

-0.70 (NGO)

-0.74 (Nobody)

(ref: government)

0.46 (Government)

-0.05 (Private)

-0.31 (NGO)

(ref: other)

n/a -0.34(Government)

0.54 (government)

-0.35 (private)

0.64** (community)

0.69 (NGO)

0.10 (other)

(ref: nobody)

Community

involvement

Experienced problem with

service 0.11 -0.37*** -0.17 0.18** 0.11 (number of problems)

Know of way to make a

complaint n/a 0.08 0.03 0.01

0.08 (fraction of services in

which a complaint was made)

Attended meeting on any

service n/a

0.00 (Knew of meeting)

0.02 (Attended meeting)

-0.21 (know of any

meetings)

0.39 (number of

meetings attended)

-0.04 (knew of meetings)

-0.01 (number of meetings

attended)

0.27*** (number of meetings)

104

Explanatory variable

DRC (Only in

certain areas)

Nepal (Never/ Almost

never) Pakistan Sri Lanka

Uganda (Almost

never)

Was consulted around any

service n/a 0.24** 0.75 -0.31***

-0.27**(number of

consultations)

Information about

the regression model

R2 0.11 0.13 n/a 0.18 n/a

Number of observations 526 1312 2088 1182 1351

Table 19: Comparative analysis: Central government decisions reflect respondent’s priorities (Very much/ to a large extent or Always).

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan (Largely or

completely)

Sri Lanka

(Completely)

Uganda (To a large

extent/ Always)

Household

factors

Female respondent -0.30 0.02 0.33 0.26 (male) 0.12

Age of respondent 0.01 -0.01 0.01 0.01 0.00

Main activity of household -0.35 (main income source:

agriculture)

1.04** (main income source:

own cultivation)

-0.64 (farming main activity of

head)

-14.84 (casual labour main

activity of head)

-0.21 (casual labour, non-agri,

main activity of head)

-1.05 (own business main

activity of head)

-0.95 (head has no activity)

-0.19 (respondent is

employed)

Respondent livelihood:

-0.33 (casual labour: agr)

-0.41 (casual labour: non-agr)

-1.17** (exploitation of bush

products)

0.58 (no activity)

0.19 (other economic activity)

0.27 (own business: shop

building)

0.73 (business: home,

market)

0.17 (fishing)

-0.29 (livestock)

0.09 (paid housework)

-2.14 (private/ NGO)

0.17 (work for government)

(ref: none)

Education level of

respondent -0.11 (categorical) -0.07 (categorical)

-0.80* (head has some

education)

0.33 (respondent completed

primary)

Respondent educ:

0.08 (some primary)

0.10 (primary)

-0.60 (O level)

-0.08 (A level)

-0.08 (tertiary)

(ref: none)

Household has migrant n/a -1.23 -0.88 -0.08 -0.15

Household receives

remittances 0.98* 0.8 0.28 -0.05 -19.40***

105

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan (Largely or

completely)

Sri Lanka

(Completely)

Uganda (To a large

extent/ Always)

Household has been

displaced

-0.84 (displaced due to

conflict)

0.03 (Lived in village whole

life)

14.12 (displaced due to

conflict) 0.04 (ever displaced) -0.23 (ever displaced)

Dependency ratio 0.07 -0.63 -0.32 0.19 0.00

Religion -1.40 (minority) n/a n/a n/a n/a

Ethnicity 0.54 (minority) 0.31 (minority) n/a --0.38 (minority)

0.44 (Langi)

-22.65*** (Iteso)

-0.15 (Kumam)

-20.49*** (Karamojong)

-22.10*** (Mixed)

1.16 (other)

(ref: Acholi)

Morris Index 0.00 0.01** 0.02*** 0.02 0.01

Food Security Index 0.06** 0.03 0.00 -0.05** -0.02**

Context

Urban n/a -0.23 n/a 0.68** n/a

Household has

experienced conflict in

past three years

-0.35 -1.13** n/a n/a -0.30

Perception of safety 0.63 (safe in village)

-0.56 (safe while travelling) -0.78 (safe in neighbourhood)

15.93 (safe when moving

around)

-1.12* (safe in

neighbourhood) n/a

Location n/a

0.62 (Rolpa)

0.47 (Bardiya)

(ref: Ilam)

-1.09* (Swat) (ref: Lower Dir) n/a -0.77 (Sub-region: Langi)

(reference: Alcholi)

Shocks

Shocks experienced 0.05 -0.13 0.41* 0.03 0.04*

Crimes experienced -0.35 -0.34 0.12 -0.37 0.01 (crimes)

-0.06** (serious crimes)

Access to services

Distance to health 0.00 -0.01 0.00 0.00 0.00

Distance to water 0.01 -0.02 n/a 0.01 0.00

Receives social protection 0.74* 1.04** -0.28 0.41* 0.69

Receives livelihood

transfer -0.02 -0.1 -0.37 0.03 0.09

Perception of

services Satisfied with health

service overall 0.28 (Satisfied) -0.39 (Dissatisfied)

n/a

0.26 (satisfied)

-0.96** (fairly satisfied)

-0.74* (dissatisfied)

(ref: satisfied)

106

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan (Largely or

completely)

Sri Lanka

(Completely)

Uganda (To a large

extent/ Always)

Number of staff at health

facility n/a

1.18 (satisfied)

-15.57 (dissatisfied)

(ref: neutral)

n/a n/a

-0.73*** (fairly satisfied)

-1.06*** (dissatisfied)

(ref: satisfied)

Availability of equipment/

medicine n/a

1.49 (satisfied)

2.89** (dissatisfied)

(ref: neutral)

n/a 0.03 (satisfied)

0.67* (fairly satisfied with

equipment)

0.56* (dissatisfied with

equipment)

(ref: satisfied)

Waiting time at health

facility n/a

-1.08*( satisfied)

-0.22(dissatisfied)

(ref: neutral)

n/a 0.26 (satisfied)

-0.40 (fairly satisfied)

-0.86* (dissatisfied)

(ref: satisfied)

Queue for water 0.17 0.36 n/a 0.15 0.36

Water clean and safe 0.37 -0.84 n/a -0.79** 0.24

Formal fees for health -0.73 n/a n/a 0.16 -0.01

Informal fees for health 0.21 1.46** n/a 0.57 0.66*

Fees for water 0.45 1.19* n/a -0.13 0.18

Provider of health service -0.02 (Government) 0.43 (Government) n/a 0.20 (Government)

-0.30 (private)

-0.87 (religious)

-23.19*** (NGO)

(ref: government)

Provider of water

-0.57 (Committee)

-0.14 (NGO)

-0.16 (Nobody)

(ref: government)

-0.57 (Government) n/a 0.00 (Government)

0.06 (government)

-0.69* (private)

-0.23(community)

0.16 (NGO)

(ref: nobody)

Community

involvement

Experienced problem with

service -0.08 -0.68** -0.27 -0.35*** -0.13 (number of problems)

Know of way to make a

complaint n/a 0.34*** 1.90*** 0.06

0.11 (fraction of services in

which a complaint was made)

107

Explanatory variable

DRC (For the most

part/ Always)

Nepal (Largely or

completely)

Pakistan (Largely or

completely)

Sri Lanka

(Completely)

Uganda (To a large

extent/ Always)

Attended meeting on any

service n/a

0.00 (Knew of meeting)

0.15 (Attended meeting

-17.07 (know of any meetings)

17.64 (number of meetings

attended)

-0.02 (knew of meetings)

0.07 (number of meetings

attended)

0.26** (number of meetings)

Was consulted around

any service n/a 0.40** 1.02* 0.20**

0.14 (number of

consultations)

Information about

the regression

model

R2 0.11 0.13 n/a 0.18 n/a

Number of observations 526 1312 2088 1182 1351

108

Annex 2: Descriptive tables

Table 1: Households receiving main income from subsistence farming, by country 109

Table 2: Mean number of income sources per household, by country 109

Table 3: Percentage of households with only one income source, out of all households with any income sources,

by country 109

Table 4: Main source of household income, DRC 109

Table 5: Main source of household income, Nepal 109

Table 6: Main source of household income, Pakistan 110

Table 7: Main source of household income, Sri Lanka 110

Table 8: Main source of household income, Sri Lanka 111

Table 9: Distance to health clinic (measured by time taken for a return journey), by country 111

Table 10: Distance to primary school (measured by time taken for a return journey), by country 111

Table 11: Distance to water source (measured by time taken for a return journey), by country 112

Table 12: Households receiving social protection and livelihood assistance, by country 112

Table 13: Satisfaction with health facility used 112

Table 14: Satisfaction with primary school used 112

Table 15: Perception of water quality, by country 113

Table 16: Transfer had any positive impact, by type of transfer and across countries 113

Table 17: Local government cares about my opinion 113

Table 18: Central government cares about my opinion 113

Table 19: Local government decisions reflect my priorities 114

Table 20: Central government decisions reflect my priorities 114

Table 21: Uptake of social protection, by country and specific transfer 115

Table 22; Uptake of livelihood assistance, by country and specific transfer 116

Table 23: List of shocks in survey instrument, by country 117

109

Table 1: Households receiving main income from subsistence farming, by country

Country Subsistence farming % Total

DRC 637 70.2 908

Nepal 1,461 46.0 3,173

Pakistan 305 15.7 1,946

Sri Lanka 390 30.6 1,275

Uganda 1,462 83.0 1,762

Total 4,255 46.9 9,064

Table 2: Mean number of income sources per household, by country

Country n Mean Min Max

DRC 1259 2.30 0 7

Nepal 3174 2.06 0 6

Pakistan 2114 1.84 0 5

Sri Lanka 1377 1.70 0 5

Uganda 1773 2.89 0 8

Total 9697 2.14 0 8

Table 3: Percentage of households with only one income source, out of all households with any income

sources, by country

Income diversity One source (%) More than one source (%) Total

DRC 16.2 83.8 100

Nepal 28.2 71.8 100

Pakistan 36.9 63.1 100

Sri Lanka 46.6 53.4 100

Uganda 11.0 89.0 100

Total 28.1 71.9 100

Table 4: Main source of household income, DRC

Livelihood Freq %

Subsistence farming 637 70.2

Casual labour (any) 96 10.6

Petty trade 64 7.0

Own business 17 1.9

Private sector (non-agri) 59 6.5

Remittances 9 1.0

Public sector 17 1.9

Other 9 1.0

Total 908 100.0

Note: Other includes domestic work and social protection transfers

Table 5: Main source of household income, Nepal

110

Livelihood Freq %

Subsistence farming 1,461 46.0

Casual labour (any) 515 16.2

Petty trade 95 3.0

Own business 383 12.1

Private sector (non-agri) 143 4.5

Remittances 306 9.6

Public sector 249 7.8

Other 21 0.7

Total 3,173 100

Note: Other includes domestic work and social protection transfers

Table 6: Main source of household income, Pakistan

Livelihood Freq %

Subsistence farming 305 15.7

Casual labour (any) 449 23.1

Petty trade 29 1.5

Own business 205 10.5

Private sector (non-agri) 160 8.2

Remittances 605 31.1

Public sector 137 7.0

Other 56 2.9

Total 1,946 100

Note: Other includes domestic work and social protection transfers

Table 7: Main source of household income, Sri Lanka

Livelihood Freq %

Subsistence farming 390 30.6

Casual labour (any) 471 36.9

Petty trade 0 0.0

Own business 178 14.0

Private sector (non-agri) 56 4.4

Remittances 40 3.1

Public sector 85 6.7

Other 55 4.3

Total 1,275 100

Note: Other includes domestic work and social protection transfers

111

Table 8: Main source of household income, Sri Lanka

Livelihood Freq %

Subsistence farming 1,462 83.0

Casual labour (any) 25 1.4

Petty trade 27 1.5

Own business 55 3.1

Private sector (non-agri) 71 4.0

Remittances 0 0.0

Public sector 3 0.2

Other 119 6.8

Total 1,762 100

Note: Other includes domestic work and social protection transfers

Table 9: Distance to health clinic (measured by time taken for a return journey), by country

30 minutes or less 31-60 minutes 61-120 minutes More than 2 hours Total

Freq % Freq % Freq % Freq % Freq %

DRC 630 51.7 378 31.0 168 13.8 43 3.5 1219 100

Nepal 2081 65.6 685 21.6 296 9.3 111 3.5 3173 100

Pakistan 1416 67.7 480 22.9 176 8.4 21 1.0 2093 100

Sri Lanka 943 68.6 215 15.6 164 11.9 53 3.9 1375 100

Uganda 340 19.4 461 26.3 432 24.6 520 29.7 1753 100

Total 5410 56.3 2219 23.1 1236 12.9 748 7.8 9613 100

Table 10: Distance to primary school (measured by time taken for a return journey), by country

30 minutes or less 31-60 minutes 61-120 minutes More than 2 hours Total

Freq % Freq % Freq % Freq % Freq %

DRC 427 47.6 262 29.2 164 18.3 44 4.9 897 100

Nepal 1445 82.8 231 13.2 64 3.7 6 0.3 1746 100

Pakistan 1370 93.4 92 6.3 5 0.3 0 0.0 1467 100

Sri Lanka 706 88.6 49 6.2 30 3.8 12 1.5 797 100

Uganda 454 47.6 286 30.0 164 17.2 49 5.1 953 100

Total 4402 75.1 920 15.7 427 7.3 111 1.9 5860 100

112

Table 11: Distance to water source (measured by time taken for a return journey), by country

30 minutes or less 31-60 minutes 61-120 minutes More than 2 hours Total

Freq % Freq % Freq % Freq % Freq %

DRC 911 75.7 224 18.6 63 5.2 6 0.5 1204 100

Nepal 1058 93.6 62 5.5 10 0.9 1 0.1 1131 100

Pakistan 721 93.8 43 5.6 5 0.7 0 0.0 769 100

Sri Lanka 538 93.6 31 5.4 5 0.9 1 0.2 575 100

Uganda 1033 59.1 468 26.8 187 10.7 61 3.5 1749 100

Total 4261 78.5 828 15.3 270 5.0 69 1.3 5428 100

Table 12: Households receiving social protection and livelihood assistance, by country

Social protection Livelihood assistance Total

Freq % Freq % Freq %

DRC 280 22.2 208 16.5 1259 100

Nepal 1205 38.0 519 16.3 3174 100

Pakistan 519 24.6 508 24.0 2114 100

Sri Lanka 356 25.9 445 32.4 1375 100

Uganda 69 3.9 273 15.4 1773 100

Total 2429 25.1 1953 20.1 9695 100

Table 13: Satisfaction with health facility used

Country (***)

Health facilities DRC Nepal Pakistan Sri Lanka Uganda All

Very dissatisfied 71 40 92 0 0 203

Dissatisfied 156 152 399 538 371 1,616

Neutral/ Fairly satisfied 393 653 145 0 987 2,178

Satisfied 373 1,958 1,203 814 354 4,702

Very satisfied 29 345 272 0 0 646

Total 1,022 3,148 2,111 1,352 1,712 9,345

Note: Significance based on Chi-squared test. *** p<0.01 ** p<0.05 * p<0.1

Table 14: Satisfaction with primary school used

Country (***)

Education DRC Nepal Pakistan girls Pakistan boys Sri Lanka Uganda All

Very dissatisfied 50 3 30 18 0 0 101

Dissatisfied 212 31 166 150 0 335 894

Neutral/ Fairly satisfied 343 222 46 57 200 642 1,510

Satisfied 257 1,249 542 661 597 298 3,604

Very satisfied 30 236 137 237 0 0 640

Total 892 1,741 921 1,123 797 1,275 6,749

Note: Significance based on Chi-squared test. *** p<0.01 ** p<0.05 * p<0.1

113

Table 15: Perception of water quality, by country

Country (***)

Water clean and safe DRC Nepal Pakistan Sri Lanka Uganda All

Yes 873 2,758 1,975 1,262 1,321 1,281

No 262 326 139 115 439 8,189

Total 1,135 3,084 2,114 1,377 1,760 9,470

Note: Significance based on Chi-squared test. *** p<0.01 ** p<0.05 * p<0.1

Table 16: Transfer had any positive impact, by type of transfer and across countries

Social protection Livelihood assistance

Transfer had any positive impact Had an impact % Total Had an impact % Total

DRC 146 61.6 237 87 62.1 140

Nepal 246 62.3 395 181 84.6 214

Pakistan 268 65.7 408 219 43.1 508

Sri Lanka 55 21.2 259 95 76.6 124

Uganda 15 60.0 25 61 52.6 116

All 730 55.1 1,324 643 58.3 1,102

Table 17: Local government cares about my opinion

Country No % Yes % Total

DRC *** 1,087 86.3 172 13.7 1,259

Nepal *** 1,799 66.2 917 33.8 2,716

Pakistan *** 1,996 94.4 118 5.6 2,114

Sri Lanka *** 486 39.8 736 60.2 1,222

Uganda *** 1,029 60.0 687 40.0 1,716

Note: Significance based on T-test of means. *** p<0.01 ** p<0.05 * p<0.1

Table 18: Central government cares about my opinion

Country No % Yes % Total

DRC *** 1,085 86.2 174 13.8 1,259

Nepal 2,028 78.8 544 21.2 2,572

Pakistan *** 2,030 96.0 84 4.0 2,114

Sri Lanka *** 580 56.0 456 44.0 1,036

Uganda *** 1,097 64.3 609 35.7 1,706

Note: Significance based on T-test of means. *** p<0.01 ** p<0.05 * p<0.1

114

Table 19: Local government decisions reflect my priorities

Country (***)

Government decisions

reflect my priorities DRC Nepal Pakistan Sri Lanka Uganda

Never 725 1,585 1,911 137 697

Almost never 0 268 79 172 324

In some areas 154 842 75 548 468

To a large extent 90 94 43 246 168

Always/ completely 0 6 6 224 45

Total 969 2795 2,114 1,327 1,702

Note: Significance based on Chi-squared test. *** p<0.01 ** p<0.05 * p<0.1

Table 20: Central government decisions reflect my priorities

Country (***)

Government decisions

reflect my priorities DRC Nepal Pakistan Sri Lanka Uganda

Never 544 1,824 2,005 158 791

Almost never 287 275 39 220 294

In some areas 122 460 42 623 375

To a large extent 78 54 25 138 155

Always/ completely 0 2 3 65 57

Total 1031 2,615 2,114 1,204 1672

Note: Significance based on Chi-squared test. *** p<0.01 ** p<0.05 * p<0.1

115

Table 21: Uptake of social protection, by country and specific transfer

Social protection

DRC % Nepal % Pakistan % Sri Lanka % Uganda %

A. Food aid 21 A. Old age allowance 13 A. BISP 21 A. Pension 2 A. Free food aid or free

household items 1

B. Micro-finance 2 B. Single women/widow

allowance

7 B. Zakat from Govt

fund

0 B. Old age pension 4 B. School feeding

programme 0

C. Disability grant 1 C. Sadqa / Nazar 0 C. Disability Allowance 1 C. Old age pension 1

D. Stipend for girls and Dalit

Children/Students

16 D. Grant from Baitul

Mall

0 D. Fisher pension scheme 1 D. Feeding patients in

hospitals 1

E. Mid-day meal, school

uniform, cooking oil for

children

7 E. Grant from RSPs

(e.g NRSP UC Poverty

Program) or other

NGOs

0 E. Fisheries loan scheme 2 E. Retirement pension

1

F. Cash transfers for family

whose family member

disappeared during or due to

conflict

0 F. Pension 2 F. Fisher insurance

scheme

1 F. Any other money

payment from the

government or

organisations?

1

G. Cash transfers for family

whose family was killed

during/due to conflict

0 G. Zakat from

community

1 G. Samurdhi 20

H. Scholarship to children of

those families whose family

members disappeared or

were killed due to conflict

0 H. Compensation for

Rehabilitation

(housing)

2 H. (Fisher) Housing

scheme

0

I. Other (specify) I. Provision of sanitary

facilities

0

J. Drinking water/Well

scheme

1

Note: Social protection transfers in red are the dependent variables for the regressions on satisfaction with social protection.

116

Table 22; Uptake of livelihood assistance, by country and specific transfer

Note: Livelihoods assistance transfers in red are the dependent variables for the regressions on satisfaction with livelihood assistance.

Livelihoods

DRC % Nepal % Pakistan % Sri Lanka % Uganda %

A. Seeds and

tools

distribution 12.4 A. Seeds and tools distribution 7

A. seeds and

tools distribution 12 A. Seeds and tools distribution 9

A. Seeds, fertilizer, pesticide

and tools 7

B.

Construction

materials

4

B. Seed money for revolving fund

(saving and credit) 1

B. Agricultural

extension 1 B. Extension services 3

B. Agricultural extension,

including training and

marketing 7

C. Agricultural extension 2

C. Training

provision 1 C. Fertilizer subsidy 8

C. Seed money for revolving

fund 6

D. Fertiliser voucher 1 D. Livestock 4 D. Credit in general 2

D. Non-agricultural services,

including training and

marketing 1

E. Goats and pigs for income

generation 5 E. Fruit saplings 3

E. Transport of fish harvest to the

market-retailer market or whole sale 1

E. Any other projects that are to

help you with your livelihood? 3

F.Skill enhancement trainings 5 F. Poultry 9 F. Market infrastructure 0

G. Micro-finance credit system

management 1

G. Agriculture

inputs 2 G. Ice factories 0

H. Teaching women about

mobilisation of funds in their

areas 1

H. Fodder /

vaccination for

livestock 1 H. Landing sites/anchorages 0

I. Marketing information 1

I. Fertilizer /

pesticide 7 I. Roads from-upto landing sites 0

J. Exposure visit 1 J. Other (specify) 1 J. Fisher related extension services 4

K. Farmers Field School 1 K. Skills training 1

L. Others (specify) 0 L. Financial management 1

M. Fuel subsidy 12

N. Beacon lights 1

117

Table 23: List of shocks in survey instrument, by country

DRC Nepal Pakistan Sri Lanka Uganda

Livestock/ crop disease Livestock/ crop disease Fire in house Livestock/ crop disease Disease of crops or livestock

Earthquake Earthquake Sudden health problem or

accident

Earthquake Bad weather

Flood Flood Long term health problem Flood Fire in house

Drought Drought Death of family member Drought Sudden health problem or

accident

Fire in house Fire in house Inflation and price hikes Fire in house Long term health problem

Sudden health problem or

accident

Sudden health problem or

accident

Loss of work of a household

member

Sudden health problem or

accident

Death of family member

Long term health problem Long term health problem Loss of land/ assets Long term health problem Inflation and price hikes

Death of family member Death of family member Land grabbed assets Death of family member Loss of job of a household

member

Inflation and price hikes Inflation and price hikes Failure or loss of family

business

Inflation and price hikes Land dispute

Loss of job of a household

member

Loss of job of a household

member

Low market prices for

livestock/ crops

Loss of work of a household

member

Other (specify)

Loss of land/ assets Loss of land/ assets Poor market access Loss of land/ assets

Theft of land/ assets Theft of land/ assets Loss of crop(s) / livestock Theft of land/ assets

Rustling Other (specify) Loss of Housing Other (specify)

Physical assault Soil problem/ losing fertility

Witchcraft Imprisonment

Other (specify) Lost irrigation channel /

system

Other (specify)

SLRC Working Papers present research questions, methods, analysis and discussion of research results (from case studies or desk-based research) on issues relating to livelihoods, basic services and social protection in conflict-affected situations. They are intended to stimulate debate on policy implications of research findings.

This and other SLRC reports are available from www.securelivelihoods.org. Funded by DFID, the EC and Irish Aid.

The views presented in this paper are those of the author(s) and not necessarily the views of SLRC, DFID the EC and Irish Aid. ©SLRC 2015.

Readers are encouraged to quote or reproduce material from SLRC Working Papers for their own publications. As copyright holder, SLRC requests due acknowledgement and a copy of the publication.

Secure Livelihoods Research Consortium (SLRC)Overseas Development Institute (ODI) 203 Blackfriars RoadLondon SE1 8NJUnited Kingdom

T +44 (0)20 7922 8249F +44 (0)20 7922 0399 E [email protected]